Biomimetic Chemistry of Iron, Nickel, Molybdenum, and Tungsten in Sulfur-Ligated Protein Sites

Abstract

Biomimetic inorganic chemistry has as its primary goal the synthesis of molecules that approach or achieve the structures, oxidation states, and electronic and reactivity features of native metal-containing sites of variant nuclearity. Comparison of properties of accurate analogues and these sites ideally provides insight into the influence of protein structure and environment on intrinsic properties as represented by the analogue. For polynuclear sites in particular, the goal provides a formidable challenge for, with the exception of iron-sulfur clusters, all such site structures have never been achieved and few even closely approximated by chemical synthesis. This account describes the current status of the synthetic analogue approach as applied to the mononuclear sites in certain molybdoenzymes and the polynuclear sites in hydrogenases, nitrogenase, and carbon monoxide dehydrogenases.

Synthetic Analogues

A strategy of demonstrated value in the study of protein-bound metal sites is the synthetic analogue or biomimetic approach defined by the protocol of Figure 1 (1). As developed and implemented in this laboratory, this approach has as its objective the preparation and detailed characterization of relatively small molecules that simulate or achieve the coordination sphere, composition, stereochemistry, and oxidation states of the native metal mononuclear or polynuclear site. A structural analogue allows deduction of site characteristics common to the site and itself by property comparisons. A functional analogue supports substrate transformations to products as do enzymes, although not necessarily at the same rate or with the same stereochemistry. A functional analogue is not inevitably a structural analogue, but a high-fidelity structural analogue should be a functional analogue provided a protein environment is not obligatory to reactivity. The approach is iterative in order to improve as necessary the accuracy of a site analogue. Here we describe the current status of selected biomimetic chemistry of four metals (Fe, Ni, Mo, W) in relation to proteins that are the objects of widespread contemporary interest: molybdenum and tungsten oxotransferases and hydroxylases, iron and nickel-iron hydrogenases, iron-sulfur proteins, molybdenum-copper and nickel-iron-sulfur carbon monoxide dehydrogenases, and nitrogenase. The sites in these proteins exhibit a range of metal nuclearities and have the common features of variable oxidation states and sulfur-rich coordination environments. Space limitations do not allow detailed accounts of enzyme reactions and mechanisms, protein structure, and inclusion of all meritorious results in site modeling. The emphasis is on the native sites themselves and recent synthetic inorganic chemistry directed toward meaningful analogues of those sites. More extensive accounts of the biomimetic chemistry of protein sites are available (2,3).

Figure 1.

Schematic representation of the synthetic analogue approach to metallobiomolecule active sites. The arrow on the left signifies repetition of the process leading to an improved analogue.

Molybdenum/Tungsten Oxotransferases and Hydroxylases

These enzymes catalyze overall reaction [1] in which generalized substrate X/XO is converted to product XO/X by addition or removal of one oxygen atom whose ultimate source is water (4-7). Current interpretation of function derives substantially from more than thirty crystal

$$\text{X}+{\text{H}}_2\text{O}\leftrightarrow \text{XO}+2{\text{H}}^{+}+2{\text{e}}^{-}$$ [1]

structures of enzymes that catalyze over a dozen different reactions. Molybdoenzymes are effectively organized under the Hille classification, which is partly based on the structures of oxidized (Mo^VI) active sites and includes the DMSOR,¹ SO, and XOR families (4). Active site structures in the DMSOR family (Figure 2) contain two pyranopterindithiolate(2-) cofactor ligands as in distorted trigonal prismatic {Mo^VIO(L)(S₂pd)₂} with L = O_Ser (1, DMSOR) (8), S_Cys (2, dissimilatory NiR) (9), and Se_Cys (3, FDH) (10,11). Although FDH catalyzes the process HCO₂ ↔ CO₂ + H⁺ + 2e^- and not reaction [1], it is placed in the DMSOR family because of site structural similarity. Also shown are the tungstoenzyme sites {W^IV(OH₂)(S_Cys)(S₂pd)₂} of acetylene hydratase (4) (12) and {W^VI(SH)(Se_Cys)(S₂pd)₂} of FDH (5) (13). Sites in the SO and XOR families contain one cofactor ligand (Figure 3) and include the oxidized sites {MoO₂(S_Cys)(S₂pd)} of chicken liver and plant SO (6) (14,15), {MoO₂S(S₂pd)} of quinoline 2-oxidoreductase (7) (16), and a substrate complex of the milk XOR site (8) (17). The wealth of structural information from crystallography and X-ray absorption spectroscopy is the single most important factor leading to the development of active site analogues. The cofactor ligand functions in the ene-1,2-dithiolate form (Figure 2), the terminal reduced member of the dithiolene class of prototypical non-innocent ligands that encompasses the oxidation states (R₂C₂S₂)^2-,1-,0 (18). Biomimetic research focuses on the chemistry of mononuclear complexes in the physiological oxidation states M^IV,V,VI.

Figure 2.

Oxidized active site structures in the molybdoenzyme DMSOR family: 1, DMSO reductase (R. sphaeroides, 1.3 Å resolution, PDB code 1EU1, one conformation) (8); 2, dissimilatory nitrate reductase (D. desulfuricans, 1.9 Å resolution, PDB code 2NAP) (9); 3, formate dehydrogenase N (E. coli, 1.6 Å resolution, PDB code 1KQF) (10). Tungsten enzymes: 4, acetylene hydratase (P. acetylenicus, 1.26 Å resolution, PDB code 2E7Z) (12); 5: formate dehydrogenase (D. gigas, 1.8 Å resolution, PDB code 1H0H) (13). The pyranopterindithiolate cofactor ligand is depicted in the box (R = nucleotide).

Figure 3.

Oxidized active site structures in the molybdoenzyme SO and XOR families: 6, sulfite oxidase (chicken liver, 1.9 Å, PDB code 1SOX) (14); 7, quinoline-2-oxidoreductase (P. putida, 1.8 Å resolution, PDB code 1T3Q) (16); 8, reaction intermediate of xanthine oxidoreductase and 2-hydroxy-6-methylpurine, 2.3 Å, PDB code 3B9J) (17)

Of the many non-dithiolene ligand platforms employed, several have proven especially useful (Figure 4). Complexes of tridentate hydrotris(pyrazolyl)borate 9 with variable substituents R have been valuable in disclosing fundamental properties of molybdenum and tungsten in physiological oxidation states (19,20), albeit not in coordination environments closely similar to enzyme sites. As examples, species such as 10 execute the minimal oxygen atom transfer reaction [2] with disclosure of mechanistic details (21), and support proton-coupled electron transfer reactions and a catalytic cycle (22) that parallels the enzymatic cycle; another variant sustains the uncommon cis-Mo^VIOS fragment (23) found in the XOR family. Bidentate ligand 11 illustrates the concept of steric suppression of nonphysiological reaction [3], allowing reaction 2 with components 12 and 13 and

Figure 4.

Ligands and complexes in biomimetic molybdenum/tungsten chemistry: tridentate hydrotris(pyrazolyl)borate(1-) 9 and an Mo^VIO₂ complex 10 (L variable), bidentate 2-pyridyldiphenylmethanethiolate(1-) (11) and its sterically encumbered Mo^VIO₂ (12) and Mo^IVO (13) complexes, and dithiolate ligands 14.

$${\text{Mo}}^{\text{IV}}\text{O}+\text{XO}\leftrightarrow {\text{Mo}}^{\text{VI}}{\text{O}}_2+\text{X}$$ [2]

$${\text{Mo}}^{\text{IV}}\text{O}+{\text{Mo}}^{\text{VI}}{\text{O}}_2\leftrightarrow {\text{OMo}}^{\text{V}}-\text{O}-{\text{Mo}}^{\text{V}}\text{O}$$ [3]

a wide variety of substrates to proceed without complication (24). Dithiolate ligands 14 closely simulate the structural and electronic features imposed by the cofactor ligand. Three common and useful types are shown, of which 14a is most like the natural ligand. Substituent variation alters electron density at the sulfur atoms and modulates redox potentials.

Structural Analogues

Efficient syntheses afford analogues of reduced (desoxo Mo^IV) and oxidized (monooxo Mo^VI) sites in the DMSOR family (Figure 5) (20,25-28). The mdt ligand has not been isolated as a stable salt but is accessible by the indicated ligand transfer reaction to form the M = Mo/W dicarbonyl complexes 15. The labile carbonyl groups are displaced to yield the terminal oxo/sulfido complexes 16, the η²-carboxylates 17, and square pyramidal species 18, which are also obtainable from 17 with silylthio or silylseleno reagants. To assure stable mononuclear structures, complexes 18 are prepared with sterically bulky R-substituents. The scheme is completed by oxo transfer reactions with XO = Me₃NO or Ph₃AsO to give monooxo complexes 19 and by a sulfur atom transfer reaction to afford 20. These complexes can be isolated only with tungsten. As is the case here, Mo^VI complexes are often unstable to autoreduction in anionic sulfur ligand environments. Because isoligated complexes of Mo^VI/W^VI are always isostructural and practically isometric, the approach has afforded structural analogues 19 of three oxidized sites in the DMSOR family and one analogue 20 of a tungstoenzyme site (compare Figure 2). Further, complexes 18 are analogues of reduced sites, 17 (M = Mo) is a representative of the site of a membrane-bound NiR of E. coli (29), and Mo=S complex 16 models the formate-reduced site in a reinterpretation of the mechanism of E. coli FDH_H (30). Analogue complexes are unprotonated versions with M=Q rather than the M-QH groups often found in enzyme sites.

Figure 5.

Synthesis of unprotonated analogues of reduced (17, 18) and oxidized (19) active sites of three members of the DMSOR family (M = Mo; Q = O, S, Se). Complex 20 is an analogue of the oxidized site in W-FDH. Reduced analogues have square pyramidal structures and oxidized analogues and 17 adopt six-coordinate structures distorted toward trigonal prismatic geometry.

Structural analogues of oxidized sites of the SO and XOR families have been synthesized utilizing dithiolate 14c (Figure 6). A sequence of substitution reactions yields 21, an analogue of the SO site in which an hindered thiolate simulates conserved cysteinate in site 6 (31). In seeking XOR site analogues, W^VI complexes have again been used because of the redox instability of Mo^VI. Removal of oxo ligands as siloxane from tungstate precursors leads to 22, an analogue of an oxidized inactive site, and 23, an analogue of the site in oxidized active enzymes (32). Native sites are often protonated, a condition simulated for structural purposes by silylation to give 24 and 25. Protonated species have not yet been isolated. Complex 23 is a long-sought structural analogue with a single basal sulfido ligand, as in native site 7, a position maintained upon reduction by substrate to the (probable) Mo^IV state in 8.

Figure 6.

Synthetic schemes leading to structural analogues of the active sites of members of the SO (21) and XOR (23) families. Complex 22 represents an inactive site and 24 and 25 monoprotonated inactive and active sites, respectively, in the XOR family.

Accurate analogues present structures unmodified by a protein environment. Superposition of analogue (19, 21-25) and oxidized protein site structures reveal near-congruency in most cases, with weighted rms deviations in atom positions of ≤0.30 Å (31-33). Protein interactions evidently do not cause major perturbations of intrinsic structures as represented by analogue complexes in their crystal lattices.

Functional Analogues

Certain of the preceding complexes manifest reductase activity with biological substrates (Figure 7). Complexes 18 cleanly reduce substrate in atom transfer reactions analogous to those of DMSOR and trimethylamine N-oxide, nitrate, and selenate reductases (20,27,28), albeit at much slower rates. The Mo^VIO analogues of 19 were generated as reaction products but not isolated. Reactions are second order with associative transition states; activation parameters for the reduction of (CH₂)₄SO by [Mo(OPh)(mdt)₂]^1- in acetonitrile (k₂²⁹⁸ = 1.5 × 10⁴ M^-1s^-1) are typical: ΔH^‡ = 10 kcal/mol, ΔS^‡ = -39 eu. These reactions exhibit a small kinetic metal effect, k₂^W/k₂^Mo ≈ 6-30, in the reduction of constant substrate. A related effect observed with molybdenum and tungsten isoenzymes is intrinsic to the metals, and is associated with the periodic behavior of isostructural and isoelectronic couples of the two metals. These reactions simulate the oxo atom transfer step in a single turnover of an enzyme. In a functional NiR analogue system, oxo transfer reaction [5] is coupled to equilibrium reaction [4] in dichloromethane with the result that Michaelis-Menton saturation kinetics are followed (34).

Figure 7.

Schematic representation of functional reductase analogue reaction systems for S-oxides, N-oxides, nitrate, and selenate.

$${[{\text{Mo}}^{\text{IV}}(\text{SR})({\text{PPh}}_3){(\text{mnt})}_2]}^{1-}\leftrightarrow {[{\text{Mo}}^{\text{IV}}(\text{SR}){(\text{mnt})}_2]}^{1-}+{\text{PPh}}_3$$ [4]

$${[{\text{Mo}}^{\text{IV}}(\text{SR}){(\text{mnt})}_2]}^{1-}+{\text{NO}}_3^{-}\to {[{\text{Mo}}^{\text{VI}}\text{O}(\text{SR}){(\text{mnt})}_2]}^{1-}+{\text{NO}}_2^{-}$$ [5]

Molybdenum and tungsten dithiolene complexes form the basis for systems in which both accurate structural and suitably functional analogues have been realized, an ultimate but infrequently achieved goal in synthetic analogue chemistry. Functional analogues refer to oxotransferases that are reductases. However, the picture will remain incomplete pending the development of oxidase analogue systems. One such system has been briefly described in relation to arsenite oxidase (35). As yet there are no effective systems for substrates of the XOR family. These enzymes are hydroxylases that utilize Mo^VI-OH units as base-activated nucleophiles in the reaction pathway rather than metal sites as oxygen atom donors or acceptors (6).

Hydrogenases

Hydrogenases occur in bacteria, archaea, and eukarya and catalyze bidirectional reaction [6]. They consist of two main classes, the [NiFe] (36,37) and [FeFe] (37-39) hydrogenases, so designated because of their active site metal content. A subgroup of the first class is the [NiFeSe] hydrogenases which contain a selenocysteinate in place of a cysteinate ligand. A third class, [Fe] hydrogenases,

$${\text{H}}_2\leftrightarrow 2{\text{H}}^{+}+2{\text{e}}^{-}$$ [6]

contain mononuclear sites (40) and is not considered here. Synthetic approaches to [NiFe] and [FeFe] enzyme sites require construction of binuclear species and in the latter case, attachment to an Fe₄S₄ cubane-type cluster. The binuclear site and site fragment structures project an apparent attainability, which together with the importance of the problem, has resulted in extensive activity in site modeling, many aspects of which have been summarized (41-45). We truncate the subject to a small set of attractive developments.

[NiFe] Hydrogenases

A combination of X-ray crystallographic and FTIR studies has led to the formulations 26a for the oxidized site and 27a for the reduced site (Figure 8) (36,37,46,47). For each, a specific structure 26b (48) or 27b (49) is provided. In the oxidized site, X is a generalized bridging ligand, thought to be an oxygenic species as indicated by O in 26b, while in the reduced site the bridging position is vacant or perhaps occupied by hydride. Appropriate initial analogues are heterometal binuclear species containing the (RS)₂Ni^II(μ₂-SR)₂Fe^II fragment with Ni-Fe separations of 2.5-2.9 Å and distorted octahedral low-spin Fe^II coordinated to one carbonyl and two cyanide ligands. A significant portion of synthetic research has focused on mononuclear iron or nickel sites (Figure 9). Among sulfur-ligated species, 28 reproduces Fe(CN)₂(CO) coordination at the low-spin Fe^II site (50), 29 evolves dihydrogen from HCl nearly quantitatively in the overall reaction Ni^I + H⁺ → Ni^II + ½ H₂ (51), and 30 catalyzes D₂/H⁺ exchange consistent with enzymatic heterolytic cleavage of D₂ or H₂ (52). A variety of dinuclear Ni-Fe complexes are accessible by reactions of suitable mononuclear components, as in the formation of 31 (53). This molecule illustrates the feasibility of the Ni^II(μ₂-SR)₂Fe^II bridge fragment, low-spin Fe^II in the unit Fe(SR)₂(CN)₂(CO)₂, and a Ni^…Fe separation near the upper end found in enzyme sites. Complex 32 reveals stabilization of low-spin Fe^II in a site containing three strong field ligands, and contains a similar bridge fragment with exclusive sulfur coordination at Ni^II but a considerably longer Ni^…Fe distance (3.29 Å) (54). No functional dinuclear NiFe analogues have yet been devised. However, in a variation on a native metal, the Fe^II(CN)₂(CO) unit was replaced by isoelectronic (C₆Me₆)Ru^II to give a complex with the bridge Ni^II(μ₂-SR)₂Ru^II incorporating a metal that forms stable dihydrogen complexes. The complex reacts with dihydrogen in water, affording a structurally related dinuclear product augmented with the hydride bridge Ni^II(μ₂-H)Ru^II and protons released in solution (55). A plausible intermediate in the enzymatic uptake of H₂ is site 27a with hydride occupying the vacant bridge position.

Figure 8.

Representative hydrogenase active site structures: oxidized [NiFe] (26a) and a specific example (26b, D. vulgaris, 1.4 Å resolution, PDB code 1WUJ) (48); reduced [NiFe] (27a) and a specific example of reduced [NiFeSe] (27b, D. maculatum, 2.15 Å resolution, PDB code 1CC1) (49)

Figure 9.

Mononuclear complexes 28-30 and NiFe complex 32, and preparation of NiFe complex 31.

[FeFe] Hydrogenases

The active sites of these enzymes are termed H-clusters and exist in three oxidation levels, air-oxidized inactive (Fe^IIFe^II), oxidized-active (Fe^IIFe^I), and reduced-active (Fe^IFe^I) with proposed oxidation states indicated. The structure of an oxidized-active site 33 (Figure 10) (56) reveals a dinuclear Fe₂(μ₂-CO)(μ₂-SR)₂(CO)₂(CN) fragment containing the dithiolate SCH₂XCH₂S (in which atom/group X = CH₂, NH, or O is not definitely identified) linked by an Fe(μ₂-S_Cys)Fe bridge to an Fe₄S₄ cluster. The iron sites are six-coordinate and that distal to the cysteinate bridge contains an aquo ligand. The problem reduces to synthesis of a suitable dinuclear fragment and attachment through an unsupported bridge to the Fe₄S₄ cluster.

Figure 10.

Structure of the H-cluster of an [FeFe] hydrogenase in the oxidized active form (33, C. pasteurianum, 1.4 Å resolution, PDB code 3C8Y) (56). In the dinuclear fragment, the carbonyl bridge is slightly asymmetric (Fe-C 1.91, 1.96 Å) and the Fe-Fe distance (2.55 Å) is consistent with direct metal-metal interaction.

Approaches to the dinuclear fragment often begin with the Seyferth anion [Fe^I₂S₂(CO)₆]^2-, which is readily dialkylated to [Fe₂(SR)₂(CO)₆] (57), thereby generating a core structure closely related to that in 33. Structures with variable X and substitution of cyanide or tertiary phosphine for carbonyl have been achieved in extensive modeling studies (41,43). Carbon monoxide, cyanide, and phosphine have the common properties of π-acid behavior and stabilization of low oxidation states. In one example of structural modeling (Figure 11), dithiolate cluster 34 undergoes CO substitution to place the phosphines in 35 at the positions of cyanide in 33, followed by oxidation to yield the Fe^IIFe^I product 36 with an asymmetric carbonyl bridge (58). An approach to the entire H-cluster framework is represented by the reaction of 1:3 subsite-differentiated cluster 37 (59) with functionalized 38 to afford a product whose physicochemical properties have been interpreted in terms of the assembly 39 in which cluster and dinuclear unit are linked by an unsupported Fe(μ₂-SR)Fe bridge (60). The viability of the structure is supported by DFT calculations which indicate the Fe^IFe^I and [Fe₄S₄]²⁺ oxidation levels. The H-cluster model and certain dinuclear species are catalysts for electrochemical dihydrogen evolution at strongly reducing potentials (61) while another Fe^IFe^I complex activates dihydrogen photochemically in the form of an Fe^IIFe^II dihydride (62).

Figure 11.

Reactions leading to a models of the dinuclear fragment (36) and the entire framework (39) of the H-cluster.

Iron-Sulfur Proteins

These proteins are of widespread occurrence at all levels of life and are implicated in electron transfer and other multifarious functions (63,64). Clusters with rhomboidal Fe₂S₂ (40,41) (65,66), cuboidal Fe₃S₄ (42) (67), and cubane-type Fe₄S₄ (43) (68) cores (Figure 12) are ubiquitous and exist in at least two oxidation states necessary for redox behavior. Analogues of the 40, 43, and the {Fe(S_Cys)₄} site in rubredoxins (not shown) were prepared at the very beginning of biomimetic inorganic chemistry, and were followed later by an analogue of 42. Different core oxidation states are achieved by appropriate choice of Fe^II,III reactants in synthesis or by redox reactions of isolated clusters. The development and accomplishments of iron-sulfur analogue chemistry and a summary of synthetic methods affording iron-sulfur clusters are available (69,70). In analogue clusters, terminal cysteinate ligation is a simulated by organic thiolates. Despite the advanced state of iron-sulfur analogue chemistry, several problems, including the complicated clusters of nitrogenase and carbon monoxide dehydrogenase (vide infra) remain, while progress has been made on two others, the Rieske cluster and the fully reduced Fe₄S₄ cluster of the Fe protein of nitrogenase.

Figure 12.

Structures of the clusters {Fe₂S₂(S_Cys)₄} (40, Anabaena, 1.2-1.3 Å resolution, PDB code 1CZP) (65), {Fe₂S₂(S_Cys)₂(N_His)₂} (41, S. acidocaldarius, 1.1 Å resolution, PDB code 1JM1) (66), {Fe₃S₄(S_Cys)₃} (42, P. furious, 1.5 Å resolution, PDB code 1SJ1) (67), and {Fe₄S₄(S_Cys)₄} (43, B. thermoproteolyticus, 0.9-1.0 Å resolution, PDB code 1IQZ) (68).

Rieske proteins function in electron transport and contain cluster 41, differentiated from the more common Fe₂S₂ cluster 40 by the presence of two Im_His ligands (71). Reaction of [Fe₂S₂Cl₄]^2-with o-xylyldithiolate and the dilithium salt of 2,2′-bis(methylindolyl)phenylmethane affords the Fe^IIIFe^III complex 44 (Figure 13), having the same ligation pattern as 41 (72). It undergoes a reversible one-electron reduction at a potential ca. 150 mV less negative than [Fe₂S₂(S₂-o-xyl)₂]^2-, an accurate analogue of 40, and exhibits a rhombic EPR spectrum similar to that of a reduced (Fe^IIIFe^II) Rieske center. Potentials of proteins with cluster 40 are more negative than those of Rieske proteins, which encompass the range -150 to +400 mV vs. SHE, a variability largely due to electrostatic effects (73). Reduction of 44 apparently occurs at the FeN₂S₂ site, as for Rieske centers. Although this cluster is not an exact analogue of a Rieske cluster owing to binding of two anionic nitrogen ligands rather than neutral imidazoles, it is the closest approach yet to the native cluster itself.

Figure 13.

The oxidized Rieske cluster analogue 44 and the preparation of all-ferrous analogue 45 of the fully reduced cluster in the Fe protein of nitrogenase.

Cubane-type clusters 43 are the most widely dispersed of all iron-sulfur species and are well known in proteins and analogue molecules in the electronically delocalized core oxidation states 3+, 2+, and 1+ of redox series [7] in which adjacent members are interconverted by one-electron transfer. The proposal that the Fe protein of nitrogenase might utilize the [Fe₄S₄]^2+/1+;1+/0 couples in delivering electrons to the catalytic site of nitrogenase was followed by definite proof of the existence of the [Fe₄S₄]⁰ oxidation state in the fully reduced Fe protein (74,75). These results render an isolable

$$\begin{array}{l}\begin{array}{cccc}{[{\text{Fe}}_4{\text{S}}_4]}^0\leftrightarrow & {[{\text{Fe}}_4{\text{S}}_4]}^{1+}\leftrightarrow & {[{\text{Fe}}_4{\text{S}}_4]}^{2+}\leftrightarrow & {[{\text{Fe}}_4{\text{S}}_4]}^{3+}\end{array}\\ \begin{array}{cccc}4{\text{Fe}}^{2+}& 3{\text{Fe}}^{2+}+{\text{Fe}}^{3+}& 2{\text{Fe}}^{2+}+2{\text{Fe}}^{3+}& {\text{Fe}}^{2+}+3{\text{Fe}}^{3+}\end{array}\end{array}$$ [7]

all-ferrous analogue a significant objective. The first such species was isolated from the substitution reaction of [Fe₄S₄(PPrⁱ₃)₄]¹⁺ with cyanide in the presence of the strong reductant [Ph₂CO]^{• -} (76). However, [Fe₄S₄(CN)₄]^4-, while in the desired oxidation state, was intensely unstable to oxidation and difficult to manipulate. Use of a N-heterocyclic carbene as a strong σ-donor terminal ligand in a cluster assembly system with an Fe^II reactant and a soluble sulfide source leads to cluster 45 (Figure 13) (77), which is amenable to study. Its ⁵⁷Fe isomer shift and comparative structural parameters are fully consistent with an all-ferrous formulation. Further, the magnetic Mössbauer and integer spin EPR spectra of 45 disclose the unusual spin ground state S = 4 (78), also found for the fully reduced native Fe protein (74) where it is unique in biology. The synthetic cluster is a meaningful analogue of the protein-bound [Fe₄S₄]⁰ cluster, providing further evidence that magnetic properties arise from interactions within the core and are not strongly influenced by terminal ligation (here S_Cys vs. C). The two all-ferrous clusters complete the synthesis of isolable analogues of all known biological oxidation states of Fe₄S₄ clusters.

Carbon Monoxide Dehydrogenases

These enzymes occur in aerobic and anaerobic bacteria and archaea and catalyze the reversible interconversion of carbon monoxide and carbon dioxide in reaction [8] at two very different types of catalytic sites. Cu-Mo CODH occurs in aerobic organisms (79) while Ni-Fe CODHs are found in anaerobic organisms (80-82). The enzymes are highly significant agents in global carbon cycling. Synthetic analogue chemistry of their catalytic sites is at an early stage.

$$\text{CO}+{\text{H}}_2\text{O}\leftrightarrow {\text{CO}}_2+2{\text{H}}^{+}+2{\text{e}}^{-}$$ [8]

Mo-Cu CODH

The oxidized site 46 (Mo^VI/Cu^I, Figure 14) (83) resembles site 7 in the XOR family (Figure 3) with the inclusion of a Cu^I-S_Cys fragment bound to the basal sulfido ligand. Two synthetic approaches to the site have emerged (Figure 15). Reaction of a Cu^I triazamacrocyclic complex with a six-coordinate Mo^VOS complex 47 results in displacement of acetonitrile and formation of assembly 48 (84). This species establishes the synthetic viability of an unsupported Mo^V-S-Cu^I bridge based on an initially terminal sulfido ligand; metric features (Mo-S 2.284 Å, Cu-S 2.135 Å, Mo-S-Cu 118.9 °, Mo^…Cu 3.74 Å) resemble those of site 46. Further, EPR spectra and DFT calculations establish that unpaired electron density extends to the Cu^I site. While this complex lacks dithiolate binding simulating the cofactor ligand, it provides another example of the utility of the hydrotris(pyrazolyl)borate ligand platform in stabilizing a structural element of a molybdoenzyme site. A second relevant reaction affords the square pyramidal Mo^VI complex 49 that incorporates the enzyme site features of an apical oxo atom, basal dithiolate coordination, and basal sulfido coordination to Cu^I (85). However, it departs from 46 with two sulfido ligands which form mutually supported bridges to Cu^I. Complexes 23 and 25 (Figure 6) offer the possibility of a single basal sulfido bridge in square pyramidal coordination as in 46 but with a non-native metal. The reactivity of these species with Cu^I is currently under investigation in this laboratory.

Figure 14.

Oxidized active site structure of Cu-Mo CODH (46, O. carboxidovorans, 1.1 Å resolution, PDB code 1N5W) (83); Cu^…Mo 3.74 Å, Mo-S 2.27 Å, Cu-S 2.21 Å, Cu-S_Cys 2.22 Å, Mo-S-Cu 113°, S-Cu-S 156°. The reduced site (Mo^IV/Cu^I) has the same structure with small differences in parameters except for the Mo-S-Cu angle and the Mo^…Cu distance which increase by about 10° and 0.5 Å, respectively.

Figure 15.

Synthetic routes to bridged complexes 48 and 49 having certain structural elements in common with Mo-Cu CODH site 46.

Ni-Fe CODH

Structures of the catalytic sites, termed C-clusters and depicted generally by 50 which subsumes actual structures 51 and 52 (Figure 16), have the common features of a cubanoid NiFe₃S₄ core with an approximately planar nickel site and a tetrahedral Fe^II site bridged to the core by atom/group X and a μ₃-S atom. Construction of a site analogue can be proposed in two steps: synthesis of the NiFe₃S₄ core containing planar Ni^II, followed by binding of Fe^II to the cluster through the μ₂-S atom axial to Ni^II and a Ni-X-Fe bridge. The conversion of previously prepared cubane-type cluster 53 (86) to 54 and 55 by substitution with strong-field in-plane ligands involves spin-pairing with a concomitant tetrahedral (S = 1) → planar (S = 0) Ni^II structural change and creation of a μ₂-S atom axial to the nickel site (Figure 17) (86, 87). This accomplishes the first step. The structure of the C-cluster of CODH from C. hydrogenoformans (1.1 Å resolution) was interpreted in terms of an X = S bridge (89), prompting the introduction of a dithiolate ligand in 55 which might sustain a sulfur bridge to external Fe^II. More recently, the crystal structure of recombinant CODH maintained at fixed redox potentials revealed C-clusters of the same overall structure but with X = OH/OH₂ in 51 and X = CO₂ in 52 (Figure 16) after treatment with NaHCO₃ at a lower potential (90). This and other evidence indicates that the bridge in the active enzyme does not contain sulfide. The central feature of a structure-based mechanism involves nucleophilic attack by an Fe^II-OH group on the carbon atom in Ni^II-CO. The bridge unit Ni^II-OH/OH₂-Fe^II is unknown in any synthetic species. Consequently, the second step in creating a C-cluster remains a formidable challenge.

Figure 16.

Generalized structure 50 of the Ni-Fe CODH sites 51 and 52 (C. hydrogenoformans, 1.4-1.5 Å resolution, PDB code 3B52) (90).

Figure 17.

Synthesis of cubanoid clusters 54 and 55 by ligand substitution of 53. The 3 LS notation refers to the tridentate thiolate ligand in cluster 37 (Figure 11).

Nitrogenase

The nitrogenase complex catalyzes the six-electron reduction reaction [9] with simultaneous dihydrogen evolution, and consists of an α₂-Fe protein with an Fe₄S₄ cluster bound between subunits and an α₂β₂-MoFe protein containing two clusters in each αβ subunit (91,92). The P-cluster is

$${\text{N}}_2+6{\text{H}}^{+}+6{\text{e}}^{-}\to 2{\text{NH}}_3$$ [9]

located at the αβ interface and the cofactor cluster FeMoco is placed within the α subunit. The path of electron flow is most likely Fe₄S₄ → P-cluster → FeMoco; the substrate is bound and reduced at the cofactor cluster by a pathway yet to be defined. Synthetic attempts directed at the P-cluster (56) and FeMoco (57) (Figure 18) (93-95), the two most complicated and synthetically formidable metalloclusters in biology, commenced well before 1992 when the first crystallographic results at atomic resolution became available. The topological relationship between the two clusters is evident from comparison 58. The P-cluster contains a μ₆-S interior atom within a Fe₆(μ₂-S_Cys)₂ cavity and FeMoco an interstitial μ₆-X atom (X = C, N, O) (95) within a Fe₆(μ₂-S)₃ cage. We describe here certain encouraging developments. More extensive accounts of work in this laboratory are available (96,97).

Figure 18.

Structures of the P-cluster 56 and FeMoco 57 (A. vinelandii MoFe protein; 2.0 and 1.2 Å resolution; PDB codes 2MIN, 1M1N) of nitrogenase (93, 95) and a comparison 58 of cluster topologies.

Biomimetic cluster synthesis is based on three conceptual strategies (96,97): (i) self-assembly--self-organizing synthesis of clusters from simple mononuclear metal precursors and ligand reagents (the cornerstone approach for cluster synthesis); (ii) fragment condensation--coupling of pre-existing di- or polynuclear clusters or a cluster with a mononuclear reactant to give higher-nuclearity clusters of (in theory) predictable structures; (iii) core conversion--reorganization of a pre-existing cluster to a different core geometry by means of redox reactions, changes in ligand set, or reaction with an external reagent, usually a ligand nucleophile. All have proven useful in biomimetic research; (i) and (iii) are illustrated below.

Heterometal cubane-type clusters with [MFe₃S₄]^z cores, including those with M = Mo and V, have been prepared by procedures (i) and (ii) (97). A sequence of reactions has been devised based on these single cubane clusters leading to the all-ferrous edge-bridged double cubanes 59 (98). The clusters, which are attractive precursors because of their high nuclearity, undergo core conversion upon reaction with hydrosulfide in acetonitrile to afford the clusters 60 (99,100). These clusters feature a prominent μ₆-S interior atom, an overall connectivity pattern identical to 56, and map closely upon the native P-cluster with rms deviations in atom positions of 0.33-0.38 Å (97). The clusters 60 are first topological analogues of the P-cluster. They are not chemical models because of the presence of heterometals and μ₂-S rather than thiolate bridges. Earlier, the same topology had been achieved in larger and less tractable clusters in which P-type units were connected by sulfide bridges (101,102).

An exceptional result has been obtained by cluster self-assembly in a system containing an Fe^II precursor with strongly basic ligands, two thiols which when deprotonated function as exceptionally capacious ligands, and elemental sulfur in toluene (Figure 19) (103). Product cluster 61 (28% yield) was isolated; it was also obtained in an assembly system not requiring prior isolation of the dinuclear intermediate. A second cluster with (Me₃Si)₂N in a doubly bridging position was formed in slight amount. The toluene solvent promotes the formation of a neutral cluster in a redox-buffered reaction system containing oxidant (sulfur) and reductants (Fe^II, thiolate) which generates a product with 5Fe^II + 3Fe^III. Remarkably, 61 manifests the FeMoco topology but with μ₂-SR/R* bridges rather than sulfide and sulfide rather than X as the interstitial atom. In addition to achieving the desired topology, the results demonstrate for the first time the formation of an iron-sulfur cluster with an atom interstitial to a trigonal prismatic Fe₆ cage. Taken together, clusters 60 and 61 and the methods affording them presage continued progress toward the goal of accurate synthetic analogues of the P-cluster and FeMoco.

Figure 19.

Synthetic reactions affording topological analogues of the P-cluster (60) and cofactor cluster (61) of nitrogenase.

Prospects

A perspective on biomimetic research of metallosites as presented here is offered by an earlier commentary from this laboratory (96). “Traditionally, synthetic inorganic chemistry has provided the molecular intuition needed to interpret metal behavior in biological systems. Today, in a reversal of circumstance, metallobiomolecules are posing fundamental inorganic questions whose answers lie outside our existing knowledge. It is a fitting symmetry that the relationship between inorganic chemistry and biology has come full circle. Equally important, these [synthetic] efforts have also contributed broadly to basic areas of inorganic chemistry that would probably not have been examined otherwise; the intrinsic value of such exploration should not be underestimated.” Given the progress thus far and the growing sophistication of biomimetic methodologies, it may be anticipated that the unachieved structural analogues of the polynuclear sites set out here will be realized. The value of analogues as structural guides or determinants (Figure 1) is diminishing as the power and productivity of crystallography and spectroscopic methods flourish.

Two additional aspects of research in this field continue apace. One is the daunting challenge of analogue systems capable of enzymatic transformations using chemical apparatus as faithful to native catalysts as possible. While an operative pathway can be won only from an enzyme itself, interrogation of analogue systems can show what is possible, among which may be the actual mechanism. It remains to be learned whether diverse functional analogues operative outside of a protein and suitable for disclosure of mechanistic information can be created. Another aspect is synthesis. While experimental protocol may not be the same, the intent differs not at all from the total synthesis of organic natural products and its contribution to biosynthetic pathways. Such work is intended not only to reach targets but also may contribute to the broad and developing area of biosynthesis of metallocenters (104) by showing what synthetic routes are feasible in the absence, and possibly in the presence, of proteins. Problems in metallocenter assembly, whether in vivo or by chemical synthesis, are at the forefront of contemporary metallobiochemistry. One need look no further than the current state of understanding of the biosynthetic pathway of FeMoco (105,106) for an inspiring example.

Abbreviations (see also Figure 3)

CODH

carbon monoxide dehydrogenase

DMSOR

dimethylsulfoxide reductase

FDH

formate dehydrogenase

FeMoco

iron-molybdenum cofactor

FTIR

Fourier transform infrared

Im

imidazole

L

ligand (generalized)

M

metal (generalized)

Me₃tacn

N,N′,N′″-trimethyl-1,4,7-triazacyclononane

NiR

nitrate reductase

Q

oxygen, sulfur, selenium

S₂-o-xyl

o-xylyl-α,α'-dithiolate(2-)

S₂pd

pyranopterindithiolate(2-)

SO

sulfite oxidase

XOR

xanthine oxidoreductase