Rethinking the Nitrogenase Mechanism: Activating the Active Site

Abstract

Metalloenzymes called nitrogenases (N₂ases) harness the reactivity of transition metals to reduce N₂ to NH₃. Specifically, N₂ases feature a multimetallic active site, called a cofactor, which binds and reduces N₂. The seven Fe centers and one additional metal center (Mo, V, or Fe) that make up the cofactor are all potential substrate binding sites. Unraveling the mechanism by which the cofactor binds N₂ and reduces N₂ to NH₃ represents a multifaceted challenge because cofactor activation is required for N₂ binding and functionalization to NH₃. Despite decades of fascinating contributions, the nature of N₂ binding to the active site and the structure of the activated cofactor remain unknown. Herein, we discuss the challenges associated with N₂ reduction and how transition metal complexes facilitate N₂ functionalization by coordinating N₂. We also review the activation and/or reaction mechanisms reported for small molecule catalysts and the Haber-Bosch catalyst and discuss their potential relevance to biological N₂ fixation. Finally, we survey what is known about the mechanism of N₂ase and highlight recent X-ray crystallographic studies supporting Fe-S bond cleavage at the active site to generate reactive Fe centers as a potential, underexplored route for cofactor activation. We propose that structural rearrangements, beyond electron and proton transfers, are key in generating the catalytically active state(s) of the cofactor. Understanding the mechanism of activation will be key to understanding N₂ binding and reduction.

Graphical Abstract

The mechanism of dinitrogen reduction by nitrogenase remains enigmatic. No evidence for any partially reduced N-N bonded species has been obtained that is diagnostic for a mechanism involving the progressive reduction of the N-N bond order by “H” functionalization. The as-isolated form of the active site cofactor does not react with substrate, suggesting that it must be activated during turnover. Characterization of the structural rearrangements constituting cofactor activation will be key to unveiling the mechanism of nitrogenase action.

Introduction

Approximately eighty percent of the Earth’s atmosphere consists of chemically inert N₂ gas. Under certain conditions, most notably binding to a transition metal center, the reactivity of N₂ can be promoted. Indeed, the two reactions that provide Earth with almost all of its bioavailable nitrogen, the industrial Haber-Bosch process and biological N₂ fixation by N₂ases, take advantage of transition metal catalysis.^1–7 Both systems involve complex mechanisms not only for nitrogen reduction, but also for catalyst activation to generate metal centers sufficiently reactive to reduce N₂. In this review, we discuss how transition metals facilitate N₂ activation across a variety of catalytic systems including synthetic metal complexes, the Haber-Bosch process, and finally nitrogenase. We also discuss catalyst activation/deactivation pathways and relate these studies back to the multimetallic active site of nitrogenase which has to be activated before it can react with N₂.

I. Dinitrogen Activation with Transition Metals

Why is N₂ so hard to fix?

Dinitrogen’s chemical inertness is manifested in its molecular properties including its low proton affinity, low electron affinity, high bond dissociation energy, and low polarity.^8,9 These properties can be altered upon binding to a transition metal center (Figure 1).^10–13 For example, the first M-N₂ complex 1 (Figure 1) exhibited a 200 cm⁻¹ shift of the ν(N-N) vibration to lower frequency in its infrared (IR) spectrum vs. free N₂ (2331 cm⁻¹ as measured by Raman spectroscopy).¹⁰ This shift to lower frequency can be attributed to the donation of electron density from the occupied metal d orbitals to the vacant π*-orbitals of the N₂ ligand.¹⁴ This interaction is referred to as π-backbonding, and was first observed in CO and olefin ligated metal complexes.¹⁵ Overall, π-backbonding results in the lengthening (i.e. weakening) of the N-N triple bond (Figure 1, enclosed).

Figure 1. N₂ Coordination Modes Observed in Transition Metal Complexes.

Dinitrogen binding at transition metal centers involves the metal center’s empty d-orbitals accepting electron density from one of the lone pairs of the N₂ ligand and π-back donation of the filled metal d-orbitals into the vacant π*-orbitals of the N₂ ligand (enclosed).^14,15 Select examples of reported N₂ binding modes shown in complexes 1-4.^10–13 For 2, Ar = 3,5-C₆H₃(CH₃)₂ and R = C(CD₃)₂CH₃.

N₂ Binding Modes

The first reported M-N₂ complex exhibited terminal N₂ binding, the most common coordination mode for N₂. In the presence of 2 or more metal centers, bridging N₂ binding modes are possible (Figure 1, bottom). Complex 2 exhibits end-on/end-on N₂ binding, 3 exhibits side-on/side-on N₂ binding and 4 exhibits an asymmetrical bridging N₂ coordination mode (end-on/side-on). The N₂ binding mode may be correlated with the mechanism by which it undergoes functionalization.^16,17 For example, the bridged N₂ ligand of complex 2 undergoes reductive cleavage to yield two Mo nitride complexes, while mononuclear complex 5 (Figure 2) undergoes functionalization at the distal N atom first (vide infra).¹¹ As depicted in the resonance structure in Figure 1, the distal N atom in mononuclear M-N₂ complexes can be particularly electron rich because of π-backbonding, and is often less sterically encumbered than the proximal N atom. Other bridged N₂ binding modes are shown in complexes 3 and 4. Early transition metals such as Zr were among the first to exhibit side-on N₂ binding.^12,18 Complex 3 features a side-on bound N₂ with an N-N bond distance of 1.548 Å as determined by X-ray crystallography. As a reference, free N₂ has an N-N bond distance of 1.0975 Å, while N₂H₄ has an N-N bond distance of 1.45 Å.¹⁹ Complex 4 is a rare example of the end-on/side-on N₂ coordination mode which also features a highly activated N-N bond.^13,20 Bridging N₂ binding motifs, like those observed in complexes 2-4, may be relevant to N₂ase, given the multimetallic nature of the cofactor and its propensity to bind ligands between two iron centers (vide infra).

Figure 2. Structures of Select N₂ Fixation Catalysts.

Chemdraw representation of the nitrogenase active site (FeMoco) and select examples of molecular N₂-to-NH₃ reduction catalysts 5-7.^30–32,142

N₂ Functionalization

The discovery of molybdenum and iron in nitrogenase motivated chemists to explore transition metal based N₂ reduction catalysis, including reactions in aprotic media by Vol’pin and Schneller and in protic media by Shilov.^8,21–24 Although the catalysts in these systems were ill-defined, owing to the heterogeneous nature of the reaction systems, reproducible NH₃ and/or N₂H₄ yields suggested the intermediacy of M-N₂ complexes. Synthetic chemists were particularly interested in the less common Mo identified in the biological system. For example, Chatt and Hidai reported the stoichiometric formation of NH₃ from well-defined M-N₂ complexes (M = Mo, W) upon treatment with acid.^25–29 Decades later, the first examples of molecular catalysts, 5-7, were described (Figure 2).^30–32 Nishibayashi has recently reported a similar catalyst system to his originally reported 6 in which (PNP)Mo(N)I (PNP = 2,6-bis(di-t-butylphosphinomethyl)pyridine) and related complexes can catalyze N₂-to-NH₃ reduction in the presence of H₂O and SmI₂, which facilitate proton coupled electron transfer.³³ Remarkably, the turnover frequency (TOF) of these Mo complexes (21–117 per minute) approaches the TOF reported for nitrogenase (~120 NH₃ per minute per active site).³⁴ In addition, catalysts featuring other transition metals, including Co, Ru, Os, V, and Ti have been developed.^35–39 Next, we discuss the mechanisms of catalysts 5 and 7.^30,40–44 in Section 2, we discuss an alternative mechanism proposed for the (PNP)Mo(N)I complex 6.

N₂-to-NH₃ Catalysis with Transition Metal Complexes

The first example of a molecular N₂-to-NH₃ reduction catalyst was reported by Schrock in 2003 (Figure 2, complex 5).³⁰ Since 2003, Schrock and coworkers have reported extensive mechanistic studies, which point to a distal- or Chatt-type pathway using a single molybdenum center as shown in Figure 3 (top pathway).⁴³ The key feature of the Chatt pathway is the weakening of the N-N triple bond by progressive H-functionalization of the β-nitrogen atom. On the other hand, the alternating pathway involves sequential functionalization of both the α- and β-nitrogen atoms. Chatt believed that the strong interaction between the metal center and N_α atom in discrete M=N=NH₂ complexes precluded reactivity at the α-N atom and promoted functionalization at the more electron rich β-N atom.⁴⁵ Following H-functionalization at the β-N atom, one equivalent of ammonia is liberated and a metal nitride species formed. Schrock’s system showed that a single Mo center could reduce N₂ to NH₃ through a sequence of proton and electron transfers, and that Mo in the biological system should not be ruled out as the site of N₂ reduction.

Figure 3. Abbreviated Mechanisms for N₂-to-NH₃ Reduction Featuring N₂H_x Intermediates.

Distal/Chatt (top), alternating (bottom), and hybrid (diagonal) pathways for N₂ reduction.^44,45,47 H₂ evolution from off-path M(H)₂ species shown in grey. M(H)₂ forms from unproductive M-H bond formation vs. N-H bond formation.^50,43,44

Given the presence of iron across all nitrogenase active sites, synthetic chemists investigated the capacity of mononuclear Fe complexes to stoichiometrically reduce nitrogen to ammonia.⁴⁶ In 2013, the Peters group reported the first molecular Fe catalyst for N₂ reduction.³² Based on mechanistic investigations of this system and those with related ligand platforms, both a distal and distal-to-alternating pathway were proposed (Figure 3 shows a simplified Chatt pathway).^44,45,47 Collectively, the mechanism for molecular systems 5-7 invoke initial functionalization at the N_β atom, leading to an M-N₂-H intermediate. Even in catalyst 6, only the terminal N₂ ligands are thought to be reduced during catalysis.⁴⁸ Selective N₂-to-N₂H₄ reduction catalysis using a molecular Fe complex bearing 1,2-bis(diethylphosphino)ethane ligands has been recently reported, suggesting an alternating pathway is indeed viable.⁴⁹ Very little reduction to NH₃ is reported in this system, however, and is attributed to the insolubility of [N₂H₅][OTf] (formed from an Fe-N₂H₄ intermediate) under catalytic conditions. Studies of 5-7 highlight the capacity of terminal N₂ ligands to undergo functionalization to NH₃ or N₂H₄ via N₂H_x intermediates. These types of mechanisms (with N₂H_x intermediates) are commonly discussed with respect to biological nitrogen fixation (Section 3) despite the lack of specific evidence for any pathway. Like biological N₂ fixation (vide infra), the formation of M-H species has been proposed during N₂-to-NH₃ reduction with molecular complexes (Figure 3, grey path).^50,43,44 In fact, for 7, under the originally reported conditions, freeze-quenched ⁵⁷Fe Mössbauer spectroscopic studies of a catalytic reaction revealed that the major Fe-containing species is an off-path iron hydride borohydride complex. This catalytically inactive species can be converted back to an on-path state using proton and electron equivalents.^32,44 The consumption of proton/electron equivalents for the activation process results in formal H₂ evolution, leading to an overall reduction of N₂ to NH₃ catalytic efficiency. In some iron catalyst systems,⁵¹ light can be used to induce H₂ elimination from off-path Fe(H)₂ species,^52–54 resulting in the generation of Fe-N₂ motifs, without consumption of proton/electron equivalents and enhancing the yields of NH₃.

Structure and Properties of FeMoco

The structure of FeMoco differs significantly from the structures of the known N₂ reduction catalysts portrayed in Figure 2. FeMoco has the composition [7Fe:9S:1C:1Mo]-R-homocitrate, and is coordinated to the MoFe-protein by only two protein ligands at opposite ends of the cofactor, the sidechains of Cys 275 to the apical Fe center and His 442 to the Mo center. At the center of the cofactor is a trigonal prism of 6Fe surrounding an unprecedented interstitial carbon, with 3 “belt sulfurs” (denoted in yellow in Figure 2) bridging the two partial cubanes. Great strides toward synthesizing structurally accurate models of FeMoco have been made,^55–59 but these do not yet fully mimic the cofactor, especially with respect to the trigonal prism core.

The active site cofactor is sufficiently stable that it can be extracted from nitrogenase using N-methylformamide (NMF),^60,61 and used to reconstitute nitrogenase samples lacking the cofactor, thereby restoring catalytic activity. While this isolated cofactor can catalyze the reduction of certain substrates,⁶² no conditions have been reported under which N₂ is reduced. Furthermore, crystallographic characterization of the extracted cofactor has been elusive and the speciation of the cofactor likely depends on the extraction conditions, with NMF, citrate, dithionite decomposition products, and phosphates, representing potential ligand candidates for the unsaturated metal centers generated upon extraction (apical Fe and Mo ions).^61,63 The behavior of the various cofactors (i.e. FeVco vs. FeMoco) towards NMF may also be different.⁶⁴

II. The Haber-Bosch Process: Catalyst Activation and Mechanism of N₂ Reduction

Unlike the mono- or binuclear N₂-to-NH₃ catalysts shown in Figure 2, the industrial catalyst for N₂ reduction features multiple iron centers and alkali metal cations.^2,65,66 While other transition metals, such as Os and Ru, facilitate industrial N₂ reduction to NH₃, only the Fe-based chemistry is discussed here.^2,67 The properties of the Fe surface determine catalytic efficiency. When comparing bare Fe surfaces, 111 faces are the most active towards N₂ fixation because they feature highly substituted Fe sites that neighbor seven other iron atoms.^68,69 Within this site, between two and five iron centers are proposed to interact with both N atoms of N₂ in a side-on fashion.^70–72 The presence of multiple irons facilitates dissociative chemisorption of N₂ across an Fe surface to yield surface bound N atoms (Figure 4).⁶⁵ This initial N-N cleavage step is the rate-determining step of catalysis. H₂ gas also undergoes bond rupture to form surface H atoms which recombine with the N atoms to give NH₃.

Figure 4. Simplified Reaction Mechanism for the Haber-Bosch Process.

Rate-determining chemadsorption of N₂ forms surface bound N atoms which recombine with H atoms to yield NH₃. Green circles represent a highly simplified iron surface.⁶⁵

In the presence of potassium (K), commonly added to the iron surface, an increased weakening of the N-N bond is observed.⁷¹ Indeed, both K and Al oxide additives are present in the standard industrial catalyst (the Mittasch catalyst).⁷³ These additives are called promoters because they promote catalysis by facilitating N₂ activation and maintaining the integrity of the activated catalyst.^2,68 The K + O layer serves as an electronic promoter which increases the N₂ reduction activity by facilitating N₂ dissociative chemisorption and by decreasing the concentration of NH₃ on the Fe surface.^74,75 These properties reflect the ability of the K atoms to strengthen the π-backbonding interaction between the Fe surface and N₂, thereby lowering the activation energy for N₂ dissociation.⁷¹ Al₂O₃ plays an important role as a structural promoter which helps the formation of the highly substituted iron centers upon catalyst activation (vide infra) and prolongs the lifetime of the active catalyst by stabilizing the reactive Fe sites against sintering.^68,76 In the absence of alumina, the active Fe surface inactivates more readily.^2,76 The Mittasch catalyst is activated with heat and high pressures of mixed gas containing H₂ and N₂. In the presence of nitrogen, the less active iron surfaces (100 or 110) transform into 111 surfaces.^68,77 This structural rearrangement is facilitated by Al₂O₃.⁷⁶ Interestingly, prolonged exposure of the activated surface to H₂ reverses this process, leading to a less active Fe surface.⁷⁷

Like the Haber-Bosch process, small molecule systems with two or more metal centers can cleave N₂ to form M-N species.^78,79 One notable example from the Holland group is shown in Figure 5, in which stoichiometric N₂ cleavage by β-diketiminate ligated iron complexes yields a tetrairon bis(nitride) cluster.⁸⁰ While only three irons bind the resultant N atoms in complex 9, a fourth iron center and two potassium cations play instrumental spectator roles in N₂ cleavage.^80,81 Potassium interacts with the aryl backbone of the ligands and one of the formed nitrides, while the fourth Fe center positions the potassium cations.

Figure 5. N₂ Cleavage Across Multiple Iron Centers.

An example of N₂ cleavage across multiple iron centers resulting in the formation of iron nitrides.⁸⁰

Recently, catalytic N₂ reduction by a pathway distinct from those shown in Figure 3 has been reported by Nishibayashi’s group for a Mo-based catalyst (Figure 6).⁸² Like the Haber-Bosch process, the proposed mechanism involves initial cleavage of the N-N triple bond across multiple metal centers as opposed to the progressive reduction of the N-N bond order by sequential functionalization of bound N₂ with “H”, ultimately resulting in N-N bond cleavage.

Figure 6. Dissociative Mechanism for N₂ Reduction Involving Two Metal Centers.

Generalized pathway for N₂-to-NH₃ fixation in which N-N bond cleavage occurs before N-H functionalization.⁸²

III. Introduction to the Mechanism of Biological N₂ Fixation

Nitrogenases catalyze the ATP-dependent reduction of N₂ to NH₃. The most well-studied N₂ase, Mo N₂ase, consists of two component proteins, the Fe protein and the MoFe protein. During catalysis, the two component proteins form a complex, allowing the ATP-dependent electron transfer from the Fe protein to the MoFe protein which is essential for substrate reduction. Here, we focus on the MoFe protein because it houses the active site cofactor that ultimately binds and reduces N₂. In addition to N₂, nitrogenase reduces other substrates including, but not limited to, C₂H₂, CN⁻, SCN⁻, N₃⁻ and H⁺; CO can also be reduced under certain conditions. Interestingly, with the exception of H⁺,⁸³ substrates have not been observed to bind to the as-isolated form of the cofactor, suggesting it must be transformed during the catalytic cycle into a state that can bind and reduce N₂.⁴ In the remaining two sections, we discuss hypotheses regarding the mechanism of cofactor activation and the putative nature of the elusive activated cofactor.

Early Kinetic Findings: A Foundation for N₂ase Reactivity

Key features of the kinetic mechanism of nitrogenase were established through pioneering work by Burris and coworkers, including that there is a lag phase in the appearance of H₂ (from proton reduction).⁸⁴ Through a quantitative analysis of the relationship between this lag phase and the turnover time using the Azotobacter vinelandii nitrogenase, the conclusion was reached that dissociation of the complex between the Fe protein and MoFe protein was an obligatory feature of the nitrogenase mechanism. That each step is associated with electron transfer to the MoFe-protein was subsequently demonstrated by following the EPR signal from the FeMo-cofactor during the lag period.⁸⁵ Since electron transfer and ATP hydrolysis occur during the lag phase, this indicates that ATP hydrolysis is coupled to interprotein electron transfer, but not directly to proton reduction.

The Lowe-Thorneley (LT) Model

The most detailed kinetic mechanism was developed by Lowe and Thorneley using the Klebsiella pneumoniae nitrogenase.⁸⁶ Based on these studies,^87,84,34 together with the reported reactivity profiles of mononuclear Mo and W complexes,⁴⁵ and the stoichiometry of N₂-to-NH₃ reduction requiring 6 H atom equivalents, Lowe and Thorneley developed a kinetic model for N₂ reduction and H₂ evolution at the MoFe protein (Figure 7).^86,88–91 Substrate reduction at the MoFe protein was proposed to involve multiple electron transfer cycles by the Fe protein, generating various states of the MoFe protein (E_n) where the subscript, n (n = 0–7), refers to the number of electron equivalents by which the MoFe protein has been reduced by the Fe protein relative to the as-isolated E₀ state (Figure 7, bottom left).⁸⁹ A critical feature of the LT model is that substrates do not bind to the E₀ state; instead, they are proposed to bind at E_n ≥ 2, with N₂ binding at either E₃ or E₄. Lowe and Thorneley proposed that one or both molecules of NH₃ were released from the E₅ state, or one NH₃ molecule was released from the E₅ state and the second from the E₆ or E₇ state.⁸⁹ It is important to note that, with the exception of the E₀ state, the other states were inferred from the kinetic model and could not be independently verified.

Figure 7. Simplified MoFe Protein Cycle of the Lowe-Thorneley Kinetic Model.

The various E states represent the different MoFe protein states as denoted by Lowe and Thorneley. Each arrow in the MoFe protein cycle indicates one Fe protein cycle. The enclosed activation steps represent chemical reactions required prior to N₂ binding, with N₂ binding and N₂-dependent H₂ evolution occurring at the E₃ or E₄ state and N₂-independent H₂ evolution at the E₂ state. NH₃ release occurs at E₅, E₆, and/or E₇.^86,88–91

An unstated assumption of the LT model is that the as-isolated E₀ state is an on-path intermediate en route to an activated En state competent for N₂ reduction and that the E0 state is regenerated after each N₂-to-2NH₃ cycle. Electron transfer to generate more reduced states is responsible for the observed induction period. However, the induction period could also reflect the as-isolated E₀ state being an off-path form requiring activation to yield a catalytically competent state. Indeed, other than monitoring the decrease in intensity of the E₀ signal using EPR spectroscopy, the proposed E_n states could not be directly detected so their existence was only inferred. Since the introduction of the LT model, different groups have reported the observation of various E states, including E₁H via Mössbauer spectroscopy,⁹² as well as E₂, E₄, E₇ and E₈ via various EPR spectroscopies.^93,94 Given the different assay conditions used to generate these putative E states, it is unclear whether the observed states match the E states of the LT model.

While not always stated explicitly, the LT model makes several other assumptions, some of which are discussed by Watt in a report attempting to duplicate and extend the LT kinetic model.⁹⁵ A key finding in this paper was that the published model did not quantitatively simulate the kinetic data that was used to develop the model. That the kinetic description is unreliable even early in the MoFe protein cycle is manifested by the reassignment of an intermediate originally reported as ‘E₂’ to E₃ based on the kinetics of appearance,⁹⁴ due to “uncertainties in the rate constants used in the kinetics analysis”.⁶ Another problematic feature of the LT model is the requirement that approximately half of the Fe protein be inactive to explain the measured specific activity; no evidence for this significant fraction of inactive Fe protein has ever been provided.

Other Key Mechanistic Insights: H₂/HD/N₂H₄ Formation Under N₂ Reduction Conditions

The Haber-Bosch process and synthetic model studies discussed above highlight the viability of a pathway involving complete cleavage of the N-N bond when multiple metal centers are present. On the other hand, pathways in which the N-N bond remains intact, but is progressively weakened by N-H functionalization, are commonly discussed in the context of biological nitrogen reduction even though there is no evidence (spectroscopic or crystallographic) for partially reduced N-N species en route to ammonia formation. A key piece of evidence suggesting that an alternating or distal pathway might be plausible was the detection of hydrazine during biological nitrogen reduction. Hydrazine may result from the formation of an Fe-N₂H₄ species in an alternating pathway en route to NH₃ formation (Figure 3, bottom) or be a product of acid/alkali quenching of a reduced N₂H_x species.⁸⁷ Indeed, Lowe and Thorneley proposed hydrazine was released from an E₄N₂H₂ (dinitrogen hydride) intermediate,⁸⁷ based on the fact that mononuclear M-NNH₂ complexes, which are proposed intermediates in a Chatt/distal pathway (Figure 3, top), yielded hydrazine upon acid or alkali quench.^29,96 Due to the acid/alkali work up, it is still unknown whether N₂H₄ is generated under catalytic conditions or whether it is formed upon quenching. Interestingly, N₂H₄ is a poor substrate for N₂ase.⁹⁷

Obligatory H₂ evolution is a key element of the LT model. Due to N₂-independent H₂ evolution, it is non-trivial to establish whether H₂ evolution is part of the N₂ reduction cycle. Several key experiments performed by various research groups suggested that H₂ production occurred in an N₂-dependent fashion. An important finding of Burris is that high N₂ pressures (up to 50 atmospheres) does not completely suppress H₂ formation by the Mo N₂ase, with an apparently limiting stoichiometry of one molecule of H₂ evolved per N₂ reduced;⁹⁸ however, the experimental findings cannot eliminate that the H₂ evolved is still independent of N₂ reduction (perhaps by partially inactive species that cannot reduce N₂ but can reduce H⁺). Significantly, H₂ inhibition is specific for N₂ reduction and does not affect C₂H₂ or other reduction processes.^99–101 The most convincing evidence for N₂-dependent H₂ evolution comes from isotope exchange studies. When an N₂ reduction assay was performed in the presence of D₂ and H₂O, HD was formed. Control experiments showed that N₂ was required for the formation of HD.^97,100 While CO does not inhibit H₂ evolution under proton reduction conditions by nitrogenase,⁹⁹ CO does inhibit HD formation under nitrogen fixing conditions.¹⁰⁰ Taken together, these data provided the strongest evidence that the reactivity of N₂ase towards H₂ and N₂ are intimately linked.

Both H₂ inhibition and HD formation were proposed to involve a bound diazene level intermediate also referred to as a dinitrogen hydride species by Lowe and Thorneley. The structure of this putative intermediate could not be assigned based on the data, but it was proposed that N₂ and H₂ (or D₂) could exchange at this state. Inspired by Sacco’s work on cobalt hydride complexes in which two hydride ligands could be substituted for N₂ via H₂ reductive elimination,¹⁰² Chatt proposed that the origin of the N₂-dependent H₂ evolution in biological nitrogen fixation was N₂ binding coupled to H₂ loss. In other words, N₂ binds to the cofactor by inducing reductive elimination of H₂ and displacing the H₂ ligand to form a M-N₂ species.^45,103

Spectroscopic Studies

Building on the foundations provided by Chatt, Lowe, and Thorneley, a detailed draft mechanism for the nitrogenase catalyzed reduction of N₂ has been developed by Hoffman, Seefeldt, and Dean. The most significant differences from the Lowe-Thorneley model are the introduction of an E₈ state (Lowe and Thorneley only went to E₇ in their model), and the “de-emphasis” of E₃ in the binding of N₂ in the LT cycle.⁹³ To date, spectroscopic and biochemical studies have characterized several of the intermediates proposed in the reduction of N₂. These include an intermediate with the stoichiometry of “2N2H” assigned to the E₄ state. The connectivity of the “2N2H” fragment is unknown, but is consistent with either a diazene level intermediate or dinitrogen dihydride species.⁹³ On the basis of freeze quenched EPR studies, a second form of the E₄ state containing two hydrides but no N species was also identified, leading to the proposal that reductive elimination of H₂ from this E₄(4H) state is required to generate Fe centers sufficiently electron rich for N₂ binding leading to give the E₄(2N2H) species.⁹³ Recent evidence established that E₄(4H) state can return to E₀ with the same EPR properties as the as-isolated form.¹⁰⁴ In addition to these E₄ species, Hoffman and collaborators have also identified the E₇ and E₈ states as the -NH₂ and -NH₃ bound FeMoco states respectively. It is important to note that the N₂H_y-FeMoco intermediates at the E₅ and E₆ levels, predicted to have reduced N-N bonds in a distal or alternating mechanism, have not been detected. Since the intermediates that have been observed to date are either not unique to any of the mechanisms discussed above (E₇ and E₈ containing a single N ligand) or the nature of the N-N bound ligand has not been conclusively established (E₄), the fundamental validity of a distal or alternating mechanism for nitrogenase has not yet been demonstrated.

IV. Iron-Sulfur (Fe-S) Lability in N₂ase: An Underexplored Activation Mechanism

Iron-Sulfur Proteins

The as-isolated state of the MoFe protein is unable to bind substrates, other than protons,⁸³ but ultimately reacts with N₂ and reduces it to NH₃. How the as-isolated MoFe protein becomes activated towards substrate binding is not understood, but in addition to electron and proton transfers, may involve some form of cluster rearrangement. Structural conversions of Fe-S clusters may represent a mechanism for tuning cluster properties, giving rise to multipurpose reactivity, that is sensitive to oxidation state, the presence of ligands, and metal availability.¹⁰⁵ Examples include redox-dependent rearrangements at several of the Fe centers of the nitrogenase P-cluster,^106,107 and at the catalytic cluster (C-cluster) of carbon monoxide dehydrogenase.¹⁰⁸ Cluster conversions can involve the loss of S, as in the 4Fe4S to 4Fe3S cluster conversion observed in hydrogenases,^109,110 or the loss of Fe as in the 4Fe4S to 3Fe4S cluster conversion in aconitase.¹¹¹ Intriguingly, in the latter system, complete loss of the cluster is associated with the ability of cytoplasmic aconitase to function as an iron response protein regulating gene transcription.¹¹² The presence of substrate under turnover conditions may preclude the incorporation of a μ₂-S bridge between the Ni and Fe centers in the C-cluster of carbon monoxide dehydrogenase.¹¹³ In an extreme case, an iron-sulfur cluster may serve as the S donor during the biosynthesis of lipoic acid.¹¹⁴ The common thread linking these systems is that the malleability of Fe-S bonds provides a versatile mechanism for tuning cluster properties in response to environmental changes. Following this theme, evidence for Fe-S bond lability at nitrogenase active sites has been garnered through crystallographic studies of FeMoco and FeVco. These observations support the idea that the reactivity of the belt S ligands may be key in activating the cofactor for catalysis by generating vacant coordination sites for substrate binding.

Mo N₂ase

In 1997, Sellman, based on studies with molecular complexes, proposed that FeMoco undergoes activation by cleavage of an Fe-S bond.^115,116 Almost twenty years later, the viability of this proposal was supported when the first ligand bound crystal structure of any nitrogenase enzyme was reported; namely, a CO-bound form of FeMoco (Figure 8, structure 10).¹¹⁷ In the X-ray structure, CO ligated Fe2 and Fe6 in a μ₂-fashion with the O atom of the CO ligand hydrogen bonded to the imidazole of His195. The overall structural change can therefore be summarized as the substitution of S2B for a CO ligand. Prior to crystallographic characterization of a CO-bound state, Q-band CW EPR, ENDOR, and IR spectroscopies provided valuable insights regarding the nature and number of CO-bound species generated under turnover conditions in the presence of various pressures of CO.^118–122 While previous spectroscopic studies provided the first direct evidence of a diatomic molecule bound to a metal center of FeMoco, with Fe6 implicated in substrate binding,^123–125 the possibility of S2B displacement had not been envisioned in these studies.

Figure 8. Active Site Structures of Mo and V Nitrogenases with Different Ligands in the Belt Positions.

Different ligand bound forms of FeMoco and FeVco with select hydrogen-bonding interactions shown in structures 10 and 13.^117,138 For 11, Se can also populate the S5A or S3A positions depending on the reaction conditions.¹²⁶ For 12 and 13, CO₃²⁻ bridges Fe4 and Fe5 of FeVco.^137,138

The displacement of S2B in structure 10 raised questions regarding the mechanism of S-substitution by CO, including the nature of the S-leaving group (S²⁻, HS⁻, H₂S, etc) and its fate. The versatility of the S2B site for ligand substitution was subsequently demonstrated by the observation that Se from selenocyanate (SeCN⁻) could also be selectively incorporated into this position under turnover conditions.¹²⁶ When the Se2B-labeled MoFe protein was subjected to turnover conditions in the presence of acetylene, Se migration to the S5A and S3A positions occurred (Figure 8, structure 11).¹²⁶ The occupancies of the S2B, S5A, and S3A positions were dependent on the amount of acetylene reduced. After a sufficient number of turnovers (thousands), Se exchanged from the cofactor. Furthermore, when a sample labeled with Se at the S2B position was treated with CO, the Se was not displaced from the cofactor, but rather migrated to the S5A and S3A positions, suggesting that CO does not directly displace S2B. Instead, S is likely lost from the S3A or S5A positions through a more complicated process involving rearrangements opening up the S2B site. After many turnovers, the Se is largely displaced from the cofactor, regenerating the all S-form of the cofactor. By highlighting the ability of the S2B, S3A and S5A belt sulfur positions to migrate and exchange with exogenous ligands, these studies have established the dynamic nature of the cofactor under catalytic conditions.

With respect to the origin of the S ligand(s) that eventually appear in the cofactor, there are multiple possibilities.^127,128 As a preemptive measure against O₂-contamination, nitrogenase purification is typically performed with buffers containing dithionite (Na₂S₂O₄). Virtually all in vitro studies, including the kinetic studies by Lowe and Thorneley, use Na₂S₂O₄ as the reductant. Dithionite is notorious for decomposing over time in aqueous solutions to various S-containing species.¹²⁹ In 2018, Hu and Ribbe et. al. reported that Na₂S₂O₄ or one of its decomposition products, sulfite (SO₃²⁻), could serve as a S source during the maturation of the L-cluster to the M-cluster.¹³⁰ Interestingly, sulfate (SO₄²⁻) and sulfide (S²⁻) did not support S-incorporation during maturation. The latter observation was surprising given the ability of sulfide to serve as a S source in many Fe-S cluster systems, including the 3Fe4S and 4Fe4S forms of aconitase.^{111,131–133} These studies suggest that multiple S reservoirs may be operative under in vivo or in vitro conditions depending on the reaction conditions. They also raise the possibility that S-based ligands may react with the cofactor under turnover conditions, hindering the observation and isolation of the activated cofactor forms.

V N₂ase

In the absence of Mo, alternative V and Fe nitrogenases may be expressed, featuring a VFe or FeFe protein respectively. Each system has its own Fe protein counterpart.^4,134 The three N₂ases exhibit distinct substrate reduction patterns.¹³⁵ For example, N₂H₄ is a product of V N₂ase while it is only detected upon acid or alkali quench with Mo N₂ase.^87,136 Another example includes reduction of CO to various hydrocarbon products, which occurs more efficiently at V N₂ase relative to Mo N₂ase. In 2017, Einsle et. al. reported the first X-ray crystal structure of the VFe protein (Figure 8, 12), unexpectedly revealing a ligand assigned as carbonate, CO₃²⁻, bridging Fe4 and Fe5 (instead of S3A as in the MoFe protein).¹³⁷ Even more recently, Einsle et al. reported the crystallographic characterization of a light atom bound FeVco state. The light atom bridges Fe2 and Fe6 (in the S2B position) and is believed to be a imido or hydroxyl containing ligand (Figure 8, 13), perhaps corresponding to a turnover intermediate such as the E₆ state.^138,139 While the authors favor the assignment of the light atom as an imido group, theoretical studies suggest the light atom is a hydroxyl group.¹⁴⁰ The crystal structure also revealed a potential sulfur binding site, approximately 7 Å away from the cofactor near the highly conserved Gln176 (corresponding to Gln 191 in the MoFe protein). This ‘turnover’ state was observed at a low dithionite concentration, again suggesting that dithionite could be a non-benign participant in cofactor speciation.

Summary and Conclusions

Despite decades of exciting progress and unexpected observations, the substrate reduction mechanism of nitrogenase remains enigmatic. At the highest level of description, it is clear that both the substrate and the catalytic cofactor must be activated during the catalytic cycle. While most mechanistic discussions of nitrogenase have favored a distal or alternating (Figure 3) mechanism for N₂ reduction to ammonia, to date, no direct evidence for an on-path, partially reduced N-N bonded species has been obtained that is diagnostic for these mechanisms. One of the challenges with studying nitrogenase is that the as-isolated form of the cofactor does not bind N₂ but must be activated during turnover. This activation process likely involves partial or perhaps complete dissociation of one or more belt sulfurs and subsequent rearrangement of the cofactor core, accompanied by protonation and reduction. Rearrangements may be facilitated by one of the striking features of the cofactor, namely that is it coordinated to the protein by only two residues, in stark contrast to other metalloclusters that typically have one sidechain ligand per metal. The belt sulfurs in the E₀ state appear to serve as protecting groups that can be removed in a proton- and electron-transfer coupled process, thereby deprotecting the associated Fe sites and effectively activating them to react with N₂ and other substrates. Given that the nitrogenase active site is embedded in the protein milieu in an aqueous environment, “just-in-time” generation of the activated form of the cofactor during the catalytic cycle will minimize side-reactions with the surrounding protein. While we focused on Fe-S bond dissociation in this review, another means for generating coordinatively unsaturated Fe centers is functionalization of the interstitial carbide. Indeed, Siegbahn, based on theoretical studies, has proposed that the carbide is transformed into a methyl group under catalytic conditions, liberating multiple Fe centers for N₂ binding.¹⁴¹

Overall, the (de)protection of reactive Fe centers for substrate binding at FeMoco may parallel findings in studies of molecular complexes and the Mittasch catalyst, including the fact that highly reactive catalysts are susceptible to deactivation pathways which result in off-path states. Understanding the nature of the active catalyst and potential catalyst deactivation pathways may facilitate the development of new and/or improved catalyst systems. One example discussed above includes adding alumina to the Mittasch catalyst which slows surface reconstruction to the inactive form. Another example discussed above is photo-enhanced N₂-to-NH₃ formation with molecular iron catalysts in which light is proposed to induce H₂ loss from catalytically inactive Fe(H)₂ complexes, resulting in the formation of catalytically competent Fe-N₂ complexes. The structure(s) of the activated cofactor and the pathways for deactivation to the as-isolated state remain outstanding questions. However, it does appear that dissociation of one or more of the belt S ligands is key in generating coordination sites for substrate binding. Learning how to facilitate S-dissociation and prevent cofactor deactivation could lead to enhanced reactivity.

An important parallel between the systems discussed in Sections I and II and nitrogenase is the potential to use multiple metal centers for substrate binding and reduction. The multimetallic nature of the cofactor will likely result in N₂ activation across multiple metal centers. At this time, the nature of N₂-binding cannot be precisely defined due to the paucity of experimental constraints. The power of integrating experimental constraints and computational modeling is becoming evident and will ultimately lead to a detailed understanding of the how nitrogenase reduces substrates.⁷ In the meantime, as these theoretical studies are based on the as-isolated resting state cofactor structure, experimental characterization of the activated state of the cofactor that binds N₂ and the structural rearrangements constituting cofactor activation will be critical for establishing the mechanism of biological nitrogen fixation.

Context and Scale.

Biological nitrogen fixation, the conversion of dinitrogen (N₂) to ammonia (NH₃) catalyzed by nitrogenases (N₂ases), provides Earth with ~50% of its bioavailable nitrogen. Without N₂ fixation, the chemically inert N₂ which comprises 80% of the Earth’s atmosphere would not be accessible to life. Although the reduction of N₂ to NH₃ is thermodynamically favored at ambient conditions, both the biological and industrial N₂ fixation processes have a significant energy requirement. Understanding the catalytic mechanism of N₂ase may lead to the development of new catalysts that can operate under mild conditions in water closer to equilibrium.

Most catalysts for N₂-to-NH₃ fixation, whether homogeneous, heterogeneous, or biological, contain transition metal centers that bind N₂ to lower the kinetic barrier for reduction. Exemplifying this theme, the active site cofactor of N₂ase contains eight transition metals. Although the N₂ase active site must become highly reactive during the catalytic cycle, the cofactor in the as-isolated state of N₂ase is a stable species that does not bind N₂. Consequently, cofactor activation is required prior to N₂ binding and functionalization. Herein, we discuss potential routes for cofactor activation based on recent studies of the Mo and V N₂ases. Like other transition metal systems capable of N₂ reduction, such as the Mittasch catalyst used in the Haber-Bosch process, structural rearrangements to the catalyst precursor are required to generate metal centers sufficiently reactive enough to bind N₂. Characterization of the activated state(s) of the cofactor that bind N₂, and the structural rearrangements constituting cofactor activation/deactivation processes, will be key to establishing the mechanism of biological N₂ fixation. These mechanistic insights will be important for guiding the development of new catalysts or improved reaction conditions for N₂ reduction with more favorable energetic requirements than current processes.