Incorporating light atoms into synthetic analogues of FeMoco

Nitrogen is an essential element for all life on Earth. However, the elemental form of dinitrogen (N₂) is typically inert, and must be converted to the more reactive and biologically accessible ammonia (NH₃) before incorporation into proteins, nucleic acids, and other biomolecules. In nature, the only enzymes capable of the multielectron reduction of N₂ to NH₃ are nitrogenases, whose complexity has captured the imagination of biochemists (1), synthetic chemists (2), and spectroscopists (3) alike. Their active sites are iron–sulfur clusters that are produced through elaborate biosynthetic pathways (4). One special aspect of the nitrogenase active-site clusters is the presence of molybdenum or vanadium heteroatoms in addition to iron: the cofactors are described as the iron–molybdenum cofactor (FeMoco) or the iron–vanadium cofactor (FeVco). Because of the unusual shape of these clusters, and the desire to systematically understand the influence of the Mo/V atom on an iron–sulfur cluster, chemists have long striven to synthesize simpler analogs (5). These synthetic analogs are influential because they can demonstrate feasible mechanistic steps and can enable correlation of specific structural features with spectroscopic signatures. In a relevant example, synthetic chemists have prepared eight-metal clusters with topological similarity to the FeMoco and FeVco (6–9).

However, a newer challenge for synthetic chemists is presented by the light atom in the center of the FeMoco and FeVco, which has been identified as a carbide (C^4-) that has no precedent in biology (Fig. 1) (10–12). The discovery of an interstitial carbon in the cofactors raises fundamental questions about the bonding, reactivity, and role of a central carbide within an iron–sulfur cluster, and about the mechanism through which it is installed into the cluster. However, embedding a light atom like carbon into an iron–sulfur cluster is challenging. Tatsumi and coworkers have described a series of eight-iron–seven-sulfur clusters that feature light atoms such as N (7, 13), but these ligands are in bridging sites in contrast to the core carbide of nitrogenase. Toward incorporating light atoms into cluster cores, Lee and coworkers (14) have described an iron–sulfur cubane featuring a core N atom, while Ohki and coworkers (9) have formed one cluster with an O atom in the center. These are notable synthetic achievements, but because these reactions involve self-assembly, rational design of more biologically relevant clusters based on these results is challenging. Now, in PNAS, Xu et al. (15) report a systematic method for incorporating nitrogen- and oxygen-based core atoms into iron–tungsten–sulfur clusters, in a strategy that may be transferable to the challenge of embedding a carbon donor into analogs of the FeMoco.

Fig. 1.

Template-assisted synthesis of cubanes that feature a core halide ligand (X′) that can be replaced with light atoms. (Inset) The FeMoco active site of nitrogenase. Belt Fe atoms shown in red are thought to be sites of N₂ binding and reduction. Structural features of interest for investigation using synthetic model systems are indicated with colored spheres. These include Fe, carbide, sulfide, and homocitrate-supported Mo.

The new clusters (Fig. 1) include complete cubanes and incomplete cubanes that feature a single W atom, a congener of the biologically relevant Mo. Their syntheses are based on a previously developed strategy known as template-assisted assembly, which employs trisulfido metal complexes such as (Tp*)WS₃ (Fig. 1) to direct assembly of additional Fe and S atoms into clusters (16). The (Tp*)WS₃ template stays intact in each synthesis and therefore leads to products with less structural rearrangement than observed using self-assembly methods. In previous work, Holm and colleagues (17) have shown that adding iron halide salts and sodium thiolate to trisulfido metal complexes forms incomplete cubanes or complete cubanes, and that product selectivity is controlled by reaction stoichiometry. The new approach advanced by Xu et al. (15) in PNAS employs the one-electron reductant, sodium benzophenone ketyl. The resulting heterometallic clusters incorporate a core halide ligand (Cl^– or Br^–) in a binding site otherwise occupied by thiolate in previously reported clusters (X′ in Fig. 1). This seemingly minor change opens the door to synthetic opportunities, because core halides are easily replaced with other ligands—including those that feature light atoms—through metathesis reactions. The synthetic methods devised by Xu et al. (15) in PNAS may therefore enable the synthesis of model compounds that feature a more similar coordination sphere to the FeMoco in nitrogenase.

In demonstrating the utility of these new methods for installing light atoms, Xu et al. (15) exchange the core halide atom X′ for ligands that feature light atoms like N and O, such as azide (N₃^–) and methoxide (OMe^–). Two different strategies for core halide substitution are disclosed: direct salt metathesis and oxidative metathesis (15). Their results demonstrate that while incomplete cubane clusters undergo salt metathesis, complete cubanes are inert under these conditions and oxidative metathesis must instead be used (Fig. 1). At first this two-pronged approach may seem like a limitation of the synthetic method. However, the ability to choose between the complementary strategies can be useful for guiding reaction design. For example, if an oxo (O^2-) core atom were desired, then it would be appropriate to use oxidative metathesis from the complete cubanes with addition of oxo transfer reagents (for example, N₂O). The design principles implicated by these findings should aid future investigators in pursuit of new iron–sulfur clusters featuring biologically relevant light atoms.

These heterometallic clusters are interesting examples of iron–sulfur clusters featuring light core atoms, but how similar are they to the FeMoco of nitrogenase? One apparent difference is the presence of W in place of Mo. Another discrepancy is the coordination sphere at the group 6 metal; in contrast to homocitrate and histidine ligands in the FeMoco, the W atom in the new compounds is coordinatively saturated and supported by the tridentate Tp* ligand. A future avenue toward greater biological analogy would therefore be incorporating Mo instead of W as the second transition metal. Unfortunately, use of [(Tp)MoS₅]^– as a template instead of (Tp*)WS₃ leads to incorporation of sulfide in the bridging site X′, where halides bind, deactivating the cluster toward further transformations. However, other Mo templates are known, and in particular the previously reported (^tBu₃tach)MoS₃ (^tBu₃tach = 1,3,5-tritert-butyl-1,3,5-triazacyclohexane) (16) might give analogous clusters featuring biologically relevant Mo.

Another important question raised by these results is whether the synthetic methods developed by Xu et al. (15) are applicable to incorporating a carbon-based ligand to mimic the carbide in the FeMoco. Literature precedent demonstrates that a bridging iron carbene can form upon reaction of diazoalkane with a diiron bridging halide complex, at least in one case (18). The use of diazoalkanes, for example trimethylsilyldiazomethane, may therefore afford a bridging carbene ligand through oxidative metathesis with a complete cubane. Even more biologically relevant would be substitution of bridging halide X′ with a methyl ligand, because biosynthesis of the carbide ligand in the FeMoco proceeds through methyl group transfer from SAM (S-adenosyl methionine) (12). Addition of methyl anion to an incomplete cubane may give rise to a bridging methyl ligand through salt metathesis, or alternatively addition of methyl radical could be used to install a carbon ligand through oxidative metathesis.

If a C donor can be installed, the next stage will be testing whether these cubane-type iron–sulfur clusters are amenable to assembly into higher nuclearity clusters. Previous work indicates that in the presence of a reductant, related all-sulfide clusters dimerize to form edge-bridged (bis)cubanes (17). Addition of thiolate sources then causes rearrangement to form corner-sharing bis(cubanes) with the topology of the FeMoco. Will light atom fragments in the clusters reported by Xu et al. (15) stay intact upon assembly to larger clusters? If so, what reactivity patterns are characteristic of these synthetic FeMoco analogs? The ability to introduce a carbon into synthetic iron–sulfur clusters with the topology of the FeMoco may give insight into the influence of the carbide in dinitrogen reduction, in addition to fundamental knowledge about the electronic structure of the unusual cofactor.