Ligand metathesis as rational strategy for the synthesis of cubane-type heteroleptic iron–sulfur clusters relevant to the FeMo cofactor

Significance

The biosynthesis and mechanism of action of nitrogenase, an enzyme that converts dinitrogen to ammonia under ambient conditions, are problems of prominent significance in metallobiochemistry. Because the active centers of the enzyme are metal–ligand clusters, it is feasible that they are attainable by synthesis and as such are primary goals in the field of biomimetic inorganic chemistry. Here we present a ligand metathesis strategy utilizing the periodic near-identity of molybdenum and tungsten when incorporated into analogous compounds. The approach provides a pathway for constructing heterometal heteroleptic Fe–S clusters of presumed relevance to the active site. Based on cubane-type stereochemistry, clusters have been prepared allowing alterations in structure and ligand binding, and inclusion of a light core atom.

Abstract

Molybdenum-dependent nitrogenases catalyze the transformation of dinitrogen into ammonia under ambient conditions. The active site (FeMo cofactor) is the structurally and electronically complex weak-field metal cluster [MoFe₇S₉C] built of Fe₄S₃ and MoFe₃S₃C portions connected by three sulfur bridges and containing an interstitial carbon atom centered in an Fe₆ trigonal prism. Chemical synthesis of this cluster is a major challenge in biomimetic inorganic chemistry. One synthetic approach of core ligand metathesis has been developed based on the design and synthesis of unprecedented incomplete ([(Tp*)WFe₂S₃Q₃]⁻) and complete ([(Tp*)WFe₃S₃Q₄]²⁻) cubane-type clusters containing bridging halide (Q = halide). These clusters are achieved by template-assisted assembly in the presence of sodium benzophenone ketyl reductant; products are controlled by reaction stoichiometry. Incomplete cubane clusters are subject to a variety of metathesis reactions resulting in substitution of a μ₂-bridging ligand with other bridges such as N₃⁻, MeO⁻, and EtS⁻. Reactions of complete cubanes with Me₃SiN₃ and S₈ undergo a redox metathesis process and lead to core ligand displacement and formation of [(Tp*)WFe₃S₃(μ₃-Q)Cl₃]⁻ (Q = Me₃SiN²⁻, S²⁻). This work affords entry to a wide variety of heteroleptic clusters derivable from incomplete and complete cubanes; examples are provided. Among these is the cluster [(Tp*)WFe₃S₃(μ₃-NSiMe₃)Cl₃]⁻, one of the very few instances of a synthetic Fe–S cluster containing a light atom (C, N, O) in the core, which constitutes a close mimic of the [MoFe₃S₃C] fragment in FeMo cofactor. Superposition of them and comparison of metric information disclose a clear structural relationship [Tp* = tris(3,5-dimethyl-1-pyrazolyl)hydroborate(1−)].

Molybdenum-dependent nitrogenases catalyze the conversion of dinitrogen to ammonia under ambient conditions and contain the weak-field iron–molybdenum cofactor (FeMo-co) cluster bearing the [MoFe₇S₉C] core at the catalytic site. The complex physicochemical properties of this cluster free in solution and bound in the FeMo protein component of the enzyme are slowly yielding to interpretation by a broad array of physical and theoretical methods. Additional significant information would include definition of the biosynthetic pathway of FeMo-co cluster formation and methods of its nonbiological chemical synthesis. We and others are engaged in the latter challenge, having demonstrated that many types of homometallic (Fe–S) and heterometallic (M–Fe–S) clusters are synthetically accessible (1–5).

The cofactor presents cubane-type MoFe₃S₃ and Fe₄S₃ subclusters bridged together by three µ₂-S atoms (6, 7) and an interstitial μ₆-C atom centered in a trigonal prismatic Fe₆ arrangement (8–10). Such features are absent in all known synthetic homo- and heterometallic Fe–S species. Light elements (carbon, nitrogen, oxygen) are very rarely found in synthetic cores and rational synthetic strategies to include them are decidedly scarce. Bridging interactions within the cluster core are nearly always introduced by direct synthesis, which raises significant challenges in accessing FeMo-co analogs with heteroleptic cores. The invariable iron-bound S²⁻/RS⁻ (R = alkyl or aryl) core ligands in most cases account for difficulties in incorporating light elements into the core. Therefore, it is highly desirable to develop new synthetic methodologies toward this end.

In seeking new cluster types potentially suitable for elaboration into cofactor-related species, we utilize previously unreported complete (3, 7) and incomplete (2, 4, 5) cubane-type clusters [(Tp*)WFe_2,3S₃Q_3,4]^z (Fig. 1). The tridentate ligand Tp* = tris(3,5-dimethyl-1-pyrazolyl)hydroborate(1−) blocks reaction at the tungsten site, Q is a variable bridging and/or terminal ligand, and z is the cluster charge. These clusters have been synthesized with a core ligand metathesis strategy based on halide-bridged cores constructed by template-assisted assembly. Here we utilize the general isostructural principle for isovalent M = Mo and W molecules (11) recently demonstrated at length with various M–Fe–S cluster types (12), and note that the molybdenum analog of 1 is unknown (13).

Fig. 1.

Depiction of the synthesis of clusters: formation of incomplete and complete cubanes 2ab and 3ab from the mononuclear precursor 1, incomplete cubanes 4–6 by ligand substitution of 2a, and the complete cubanes 7 and 8 by redox metathesis of 3a, as well as the conversion from 2ab to 3ab (ketyl = Na⁺(Ph₂CO^•)⁻). All clusters were isolated as Et₄N⁺ salts (except for 2b as the [Fe(DMF)₆]²⁺ salt). Compounds in blue circles and 6 are incomplete cubane series and those in squares and 8 are complete cubanes. Compounds highlighted in light blue are clusters with halide bridge and those in light orange are clusters incorporating light atoms (N/O).

Results and Discussion

Clusters 2–8 are schematically depicted together with reactions initiated from precursor 1 which functions as a pyramidal WS₃ template (Fig. 1). When reduced below the W^VI oxidation level under anaerobic conditions in N,N-dimethylformamide (DMF) solutions, the template-bound Fe^{II, III} forms clusters in incomplete or complete cubane modes, which are of considerable interest in the assembly of species relevant to the native cluster. Sulfur atoms derive entirely from 1. Precursors 2 and 3 are readily formed (48–62%) by controlling stoichiometries of reactions (Eqs. 1–3; Q = Cl, Br) with sodium benzophenone ketyl as reductant. Incomplete and complete cubane clusters 4–8 with heteroligated cores were obtained (77–92%) by ligand metathesis reactions based on clusters 2 and 3. Cluster structures 2–5 and 7 were determined by X-ray single-crystal diffraction. Crystallographic data and metric parameters are collected in Supporting Information. Structures are set out in Figs. 2–4, including selected data on ligand binding.

Fig. 2.

Structures of [(Tp*)WFe₂S₃(μ₂-Q)Q₂]⁻ [Q = Cl (2a), Br (2b)] and [(Tp*)WFe₃S₃(μ₃-Q)Q₃]²⁻ [Q = Cl (3a), Br (3b)]. In this and other figures atoms are shown at 50% thermal ellipsoids probability and hydrogen atoms are omitted for clarity. Selected bond metrics (mean, Å or deg): 2a, Fe-Cl1 2.365(3), Fe1-Cl1-Fe2 67.57(7); 2b, Fe-Br1 2.518(2), Fe1-Br1-Fe2 62.83(4); 3a, Fe-Cl1 2.495(3), Fe-Cl1-Fe 63.27(9); 3b, Fe-Br1 2.622(2), Fe-Br1-Fe 59.75(6).

Fig. 4.

Structure of [(Tp*)WFe₃S₃(μ₃-NSiMe₃)Cl₃]⁻ (7). Selected metric features (mean, Å, deg): Fe-N1.936(3), Si-N 1.749(4), Fe-Cl 2.221(2), Fe-N-Si 129.1(2).

The strategy of template-assisted assembly utilized by Majumdar and Holm (14) has demonstrated the feasibility of incorporating selenium atoms into cubane-type M–Fe–S (M = Mo or W) cores, in which bridges between iron atoms are introduced independently in addition to the M-bound sulfide ligand. We have thus designed the syntheses eliminating the usage of an additional S/Se source in the assembly process and expecting halide ligand to coordinately bridge the iron atoms in the cubane-type clusters, with [(Tp*)WS₃]⁻ as template. Clusters 2a/2b and 3a/3b have been obtained as a result. For comparison, we have also examined the possibility of using [(Tp)MoS₅]⁻ as the template (See Scheme S1 in Supporting Information for structure), but in all cases examined, the free sulfide generated during the assembly process occupies the bridging site to form the [MoFe₃S₄]^z core. Its presence in the reaction system impedes halide coordination and deactivates the template.

Incomplete cubanes 2a and 2b, of an unusual structural form, are readily recognized by vacant metal sites whereas the complete clusters 3a and 3b have the full cubane [WFe₃S₃Q] core (Q = Cl or Br) (Fig. 2). Isostructural clusters 2a and 2b are prepared by direct reaction 1 and contain one Fe–Q–Fe bridge, an unusual feature in Fe–S clusters. Complete cubanes 3a/3b were obtained by two methods, reactions 2 and 3, whose products have μ₃-bridging and terminal halide ligands. The first of these is direct synthesis with a stoichiometry intended to produce a reduced core [WFe₃S₃(μ₃-Q)]²⁺. In reaction 3, clusters 3a and 3b were obtained by conversion of preisolated clusters 2a and 2b generated in reaction 1, in 75% and 83% yield, respectively. Direct synthesis gave a better yield of complete cubane clusters compared with the total yield through the preisolation pathway.

$${\left[\left({\text{Tp}}^{\ast }\right){\text{WS}}_3\right]}^{–}+2\hspace{0.17em}{\text{FeQ}}_2+{\text{Na}}^{+}{\left({\text{Ph}}_2{\text{CO}}^{•}\right)}^{–}\to {\left[\left({\text{Tp}}^{\ast }\right){\text{WFe}}_2{\text{S}}_3\left({\mu }_2-\text{Q}\right){\text{Q}}_2\right]}^{–}+\text{NaQ}+{\text{Ph}}_2\text{CO},$$ [1]

$${\left[\left({\text{Tp}}^{\ast }\right){\text{WS}}_3\right]}^{-}+3\hspace{0.17em}{\text{FeQ}}_2+3\hspace{0.17em}{\text{Na}}^{+}{\left({\text{Ph}}_2{\text{CO}}^{•}\right)}^{-}+{\text{Q}}^{-}\to {\left[\left({\text{Tp}}^{\ast }\right){\text{WFe}}_3{\text{S}}_3\left({\mu }_3-\text{Q}\right){\text{Q}}_3\right]}^{2-}+3\hspace{0.17em}\text{NaQ}+3\hspace{0.17em}{\text{Ph}}_2\text{CO},$$ [2]

$${\left[\left({\text{Tp}}^{\ast }\right){\text{WFe}}_2{\text{S}}_3\left({\mu }_2-\text{Q}\right){\text{Q}}_2\right]}^{-}+{\text{FeQ}}_2+2\hspace{0.17em}{\text{Na}}^{+}{\left({\text{Ph}}_2{\text{CO}}^{•}\right)}^{-}+{\text{Q}}^{-}\to {\left[\left({\text{Tp}}^{\ast }\right){\text{WFe}}_3{\text{S}}_3\left({\mu }_3-\text{Q}\right){\text{Q}}_3\right]}^{2-}+2\hspace{0.17em}\text{NaQ}+2\hspace{0.17em}{\text{Ph}}_2\text{CO}.$$ [3]

Structural analysis of 2a reveals that the mean Fe–Cl (bridging) bond length of 2.365(3) Å is much longer than the average Fe–Cl (terminal) bond distances at 2.230(3) Å. And in 3a, the mean Fe–Cl (bridging) contact of 2.495(3) Å is significantly longer compared with average Fe–Cl (terminal) distances at 2.284(4) Å. The longer Fe–Cl (bridging) contacts indicate that the bridging Cl sites are likely to be more susceptible to substitution than terminal sites. Therefore, a rational ligand metathesis strategy is envisioned utilizing these halide-bridged clusters as precursors for heteroligated cores.

$${\left[\left({\text{Tp}}^{\ast }\right){\text{WFe}}_2{\text{S}}_3\left({\mu }_2-\text{Cl}\right){\text{Cl}}_2\right]}^{-}+\text{NaQ}\to {\left[\left({\text{Tp}}^{\ast }\right){\text{WFe}}_2{\text{S}}_3\left({\mu }_2-\text{Q}\right){\text{Cl}}_2\right]}^{-}+\text{NaCl},$$ [4]

$${\left[\left({\text{Tp}}^{\ast }\right){\text{WFe}}_3{\text{S}}_3\left({\mu }_3-\text{Cl}\right){\text{Cl}}_3\right]}_2+{\text{Me}}_3{\text{SiN}}_3\to {\left[\left({\text{Tp}}^{\ast }\right){\text{WFe}}_3{\text{S}}_3\left({\mu }_3{-\text{NSiMe}}_3\right){\text{Cl}}_3\right]}^{-}+{\text{N}}_2+{\text{Cl}}^{-},$$ [5]

$${\left[\left({\text{Tp}}^{\ast }\right){\text{WFe}}_3{\text{S}}_3\left({\mu }_3-\text{Cl}\right){\text{Cl}}_3\right]}^{2-}+1/8\hspace{0.17em}{\text{S}}_8\to {\left[\left({\text{Tp}}^{\ast }\right){\text{WFe}}_3{\text{S}}_4{\text{Cl}}_3\right]}^{-}+{\text{Cl}}^{-}.$$ [6]

The metathesis strategy was first examined over the incomplete cubane-type cluster [(Tp*)WFe₂S₃(μ₂-Cl)Cl₂]⁻, resulting in displacement of the μ₂-bridge from chloride (2a) to azide (4), MeO⁻ (5), and EtS⁻ (6) by suitable variation of ligands Q⁻ in reaction 4 (Figs. 1 and 3). When the metathesis strategy is applied to the complete cubane cluster [(Tp*)WFe₃S₃(μ₃-Cl)Cl₃]²⁻ (3a), these ligands failed in direct substitution reactions. Usage of oxidative reactants Me₃SiN₃ and S₈ triggers redox metathesis reactions 5 and 6 to form [(Tp*)WFe₃S₃(μ₃-Q)Cl₃]⁻ [Q = Me₃SiN²⁻ (7) and S²⁻ (8); Figs. 1 and 4]. A favorable feature of these reactions is high yield. As one example, equimolar Me₃SiN₃ converts 3a to cluster 7 with the complete [WFe₃S₃(μ₃-N)] cubane core and retention of terminal chloride in 92% yield. The metathesis strategy should be amenable to extension with a variety of ligand types, including formation of the heterobridges Fe-(μ_2,3-Q)-Fe.

Fig. 3.

Structures of [(Tp*)WFe₂S₃(μ₂-Q)Cl₂]⁻ [Q = N₃ (4), OMe (5)]. Selected metric features (mean, Å, deg): 4, Fe-N1 2.045(8), N1-N2 1.079(9), N2-N3 1.22(1), Fe1-N1-Fe2 80.2(3), N1-N2-N3 178.8(8); 5, Fe1-O1 1.964(4), O1-C1 1.381(9), Fe1-O1-Fe1ⁱ 84.3(2). Symmetry code: i = +x, −y + 1/2, +z.

The [WFe₃S₃N] core in 7 demonstrates a highly close topological mimicking of the [MoFe₃S₃C] fragment of FeMo-co, as shown by least-squares structural superposition of them with very good agreement (Fig. 5). The mean Fe-N/C distances are 1.963(3)/2.00 Å and the root-mean-square deviation of fit in core atom positions is 0.0535. The fitting procedure for FeMo-co and another synthetic cubane give nearly the same result (4, 15).

Fig. 5.

X-ray structural superposition of the [WFe₃S₃N] core in 7 (blue lines) and the [MoFe₃S₃C] portion of FeMo-co (red lines). Central atom X = N or C. For further details on this type of structural comparison, see refs.4 and 15.

In assessing these reactions, it is observed that only a very limited number of light-atom bridging ligands C, N, O have been incorporated into Fe–S or M–Fe–S cubane-type structures (4). Bridging interactions nearly always are supported by μ_2–4-S connectivities formed in direct synthesis. Other than clusters originating in this work, the heteroligand μ₃-RN²⁻ is known only in the cubane series [Fe₄(N^tBu)_nS_4-nCl₄]^z whose members are obtained from condensation reactions of Fe^III amides (15). In very recent work, we have prepared additional heteroligand and heterometal clusters with nuclearities above those of individual cubanes. In view of this development and the proposed biosynthetic modular pathways of clusters leading to FeMo-co and the P^N-type cluster of nitrogenase (16–18), chemical synthesis is a goal of increasing import in a full interpretation of this enzyme.

Materials and Methods

All reactions and manipulations were performed on standard Schlenk lines or in glove box under the atmosphere of dry dinitrogen. Physical measurements and characterizations were also carried out under the protection of dry dinitrogen. All reagents used were anhydrous either commercially available or purified by conventional methods. See Supporting Information for details.