Reduction of Substrates by Nitrogenases

Abstract

Nitrogenase is the enzyme that catalyzes biological N₂ reduction to NH₃. This enzyme achieves an impressive rate enhancement over the uncatalyzed reaction. Given the high demand for N₂ fixation to support food and chemical production and the heavy reliance of the industrial Haber–Bosch nitrogen fixation reaction on fossil fuels, there is a strong need to elucidate how nitrogenase achieves this difficult reaction under benign conditions as a means of informing the design of next generation synthetic catalysts. This Review summarizes recent progress in addressing how nitrogenase catalyzes the reduction of an array of substrates. New insights into the mechanism of N₂ and proton reduction are first considered. This is followed by a summary of recent gains in understanding the reduction of a number of other nitrogenous compounds not considered to be physiological substrates. Progress in understanding the reduction of a wide range of C-based substrates, including CO and CO₂, is also discussed, and remaining challenges in understanding nitrogenase substrate reduction are considered.

Graphical Abstract

1. INTRODUCTION

The element nitrogen (N) is a constituent of many industrially and biologically important molecules, and thus there is a high worldwide demand for usable forms of N.^1–4 The major simple molecular states of N constitute the global biogeochemical nitrogen cycle, with dinitrogen (N₂) being the most abundant form constituting 78% of the Earth’s atmosphere.^5–10 While abundant, N₂ is not generally useable directly but instead must first be “fixed” into a form that is useful for industrial or biological processes.^1,2,11,12 The reduction of N₂ to NH₃, called nitrogen fixation, is the largest contributor of fixed nitrogen into the global N cycle. This reduction reaction occurs at scale in comparable amounts through two processes, the Haber–Bosch industrial reaction^13–15 and by the action of the enzyme nitrogenase operating inside specialized microorganisms, called diazotrophs, found throughout the biosphere.^11,16–18 Nitrogenase catalyzes the N₂ reduction reaction having the following optimal (limiting) stoichiometry:^19–23

$${\text{N}}_2+8{\text{H}}^{+}+16\text{MgATP}+8{\text{e}}^{-}\to 2{\text{NH}}_3+{\text{H}}_2+16\text{MgADP}+16{\text{P}}_{\text{i}}$$ (1)

As indicated by eq 1, N₂ reduction by nitrogenase is fundamentally an electrochemical process, employing electrons and protons. In that sense, it differs qualitatively from the chemical industrial Haber–Bosch process, which employs H₂ as the source of reducing equivalents and H (eq 2):³

$$3{\text{H}}_2+{\text{N}}_2\to 2{\text{NH}}_3$$ (2)

The nitrogenase catalyzed reaction requires energy input in the form of hydrolysis of ATP, whereas 90% of the energy input for the Haber–Bosch process is used to make the H₂ reactant from methane²⁴ (eq 3)

$${\text{CH}}_4+2{\text{H}}_2\text{O}\to {\text{CO}}_2+4{\text{H}}_2$$ (3)

Finally, the nitrogenase stoichiometry (eq 1) incorporates the obligatory production of one H₂ per N₂ reduced, a long-debated idea that has been established only recently.^21,23

Enzyme-catalyzed reactions, like all catalyzed reactions, are often measured by the rate enhancement achieved over the uncatalyzed reaction.^25–27 Although nitrogenase catalyzes the reduction of N₂ to 2 NH₃ with a seemingly low rate constant of ~1 s⁻¹,^19,20,28,29 this corresponds to a large enhancement over the uncatalyzed reaction that places it as one of the most impressive known catalysts.

Research on nitrogenase over decades has endeavored to understand this complex enzyme and elucidate the mechanism by which this remarkable catalyst achieves its rate enhancement.^{20,21,23,30–37} Other reviews in this volume detail diverse aspects of how nitrogenase works, including how electron transfer occurs and the roles of ATP hydrolysis, how spectroscopy provides insights into the properties of the active site metal clusters, and how metal site cofactors are biosynthesized. In this Review, we detail how the field has advanced over the last ~6 years, with an emphasis on the recent breakthroughs in our understanding of how the active site metal cluster achieves substrate reduction.

Although the physiological substrates are protons and N₂, nitrogenases can also reduce a number of other substrates, such as acetylene (C₂H₂), carbon dioxide (CO₂), diazene (HN=NH), and hydrazine (H₂N-NH₂).^20,39,40 We also review the current understanding of the mechanism of reduction of these alternative substrates, which not only is providing new insights into the mechanism of nitrogenase, but also provides insights into possible ways nitrogenase might be used to carry out useful reductions other than N₂ fixation.

2. NITROGENASES

2.1. Enzymes

There are three known isozymes of nitrogenase: the molybdenum-dependent Mo-nitrogenase, vanadium-dependent V-nitrogenase, and iron-only Fe-nitrogenase.^20,41–49 The nitrogenases are broadly distributed throughout nature with corresponding genes necessary to produce nitrogenases predicted in 13 phyla of bacteria and one phylum of archaea.^18,50 Mo-nitrogenase is the predominant isozyme,⁵¹ but many diazotrophs carry genes for at least one, if not both, of the other isozymes.⁴⁴ Because of this distribution, the V-nitrogenase and Fe-nitrogenase are often called “alternative” nitrogenases.^41,44,52 Expression of the isozymes appears to be primarily regulated by metal availability, wherein the absence of Mo or the inability to transport Mo efficiently, one of the alternative systems is expressed.^{43,44,46,49,53}

Structures of the Mo- and V-nitrogenase, reviewed elsewhere in this issue, show them to have similar quaternary structures.^54–60 The structure of Fe-nitrogenase has not been solved, but available genetic and spectroscopic evidence indicates it is similar to the other two,^43,45 as depicted in Figure 1. All three isozymes are two-component protein systems^61,62 with the catalytic components for each being encoded by unique gene clusters. Mo-nitrogenase is encoded by nif genes,^50,63 V-nitrogenase by vnf genes,^64–66 and Fe-nitrogenase by anf genes.^66,67 The two component proteins are a catalytic component, or MFe protein (M = Mo, V, Fe) (NifDK, VnfDGK, AnfDGK), which receives electrons from the electron delivery component, or Fe protein (NifH, VnfH, AnfH).^{20,31,41,43,44} The MFe proteins form as two identical catalytic halves; MoFe protein is an α₂β₂ heterotetramer and VFe and FeFe proteins are α₂β₂γ₂ heterohexamers.^41,43,68 Each catalytic half houses an [8Fe-7S] P-cluster and an active site cofactor called FeMo-co,^{55–57,69–71} FeV-co,^58,72,73 or FeFe-co³⁸ as the site of substrate binding and reduction. Fe protein for all three isozymes is a homodimer, which houses a single [4Fe-4S] cluster and two nucleotide-binding sites for ATP (Figure 1).^54,60

Figure 1.

<j/>General nitrogenase architecture. Schematic of nitrogenase structures showing the electron delivery Fe protein component and the catalytic MFe protein component (M = Mo, V, Fe). MoFe protein is an α₂β₂, and VFe and FeFe proteins are α₂β₂γ₂ heterohexamers. Adapted with permission from ref 38. Copyright 2018 American Chemical Society.

In all three isozymes, electron flow originates from the [4Fe-4S] cluster of the Fe protein, and proceeds through the P-cluster, eventually ending on the active site cofactor FeM-co (Figure 2).^74–77 The process of electron flow is complex and is covered in detail in other chapters of this Review.

Figure 2.

Structures of the FeMo-cofactor of Mo-nitrogenase,^56,57,69,79 FeV-cofactor of V-nitrogenase,^58,72 and proposed structure of the FeFe-cofactor of Fe-nitrogenase.⁸⁰ View is looking down on the Fe2, 3, 6, 7 face as indicated by the Fe atom numbering on FeMo-cofactor. Not shown are the cysteine coordinating Fe1, and the histidine and the R-homocitrate tail that ligate the Mo, V, or Fe are shown in blue.

2.2. Cofactors

Electrons and protons are accumulated, and substrates are reduced on the active site cofactor (FeMo-co, FeV-co, and FeFe-co, Figure 2). FeMo-co has the chemical composition [7Fe-9S-Mo-C-(R)-homocitrate] and is often referred to as the M-cluster.^{56,57,69,70,78,79} FeV-co has a chemical composition of [7Fe-8S-V-C-(R)-homocitrate], with a carbonate ligand bound in place of the ninth sulfur and is sometimes referred to as V-cluster.^58,59,72 The FeFe-co is less well studied, but from spectroscopic studies is inferred to have a chemical composition of [8Fe-9S-C-(R)-homocitrate].⁸⁰

3. MECHANISM OF NITROGENASE REDUCTION OF PROTONS AND N₂

Extensive studies in the 1970s and 1980s^20,31,81 led to the development of the Lowe–Thorneley kinetic scheme for catalysis by the Mo-dependent nitrogenase, Figure 3.^20,31,81 This describes the transformations among catalytic intermediates, denoted E_n where n is the number of steps of electron/proton delivery from the nitrogenase Fe protein to the active-site iron–molybdenum cofactor ([7Fe-9S-Mo-C-R-homocitrate]; FeMo-co) Figure 2.^23,82 This scheme’s central feature was the mysterious proposal of obligatory formation of one mole of H₂ per mole of N₂ reduced, with the attendant limiting stoichiometric requirement of 8[e⁻/H⁺] per reduction of one N₂ to 2NH₃, as given by eq 1, in agreement with stoichiometric measurements by Simpson and Burris,¹⁹ but in contrast to the six reducing equivalents from 3H₂ used in the Haber–Bosch reaction.¹⁴

Figure 3.

Simplified 8[e⁻/H⁺] kinetic scheme for nitrogen reduction. In the Lowe–Thorneley E_n notation, n = number of [e⁻/H⁺] added to FeMo-co; in parentheses, the stoichiometry of H/N bound to FeMo-co. The re/oa equilibrium is highlighted in red, and the intermediates in red boxes have been freeze trapped for spectroscopic study.

Some early reports presented possible explanations for the proposed connection between H₂ evolution and N₂ reduction, including models that involve metal-bound hydrides.^{20,31,83–92} These speculations invoked multiple forms of hydride intermediates,^31,88,90 as they sought to explain three important observations of nitrogenase catalysis: (i) obligatory H₂ evolution during N₂ reduction by Mo-nitrogenase (eq 1),^19,93 (ii) H₂ (D₂) as competitive inhibitors of N₂ reduction,^88,90,94,95 and (iii) N₂-dependent HD formation by Mo-nitrogenase during N₂ reduction in the presence of D₂.^{85,88–90,96–100} One report suggested a dihydride intermediate after accumulation of three or four reducing equivalents (E₃ or E₄ in Figure 3) on the active site, and even suggested that N₂ somehow binds by displacing one H₂.^31,87 Metal-hydride intermediates were also proposed to be involved in the reduction of C-based substrates (e.g., C₂H₂ and cyclopropene).^101–103 In parallel, progress in understanding binding and activation of small molecules, such as N₂ and NO₃⁻, by metal-hydride and metal-dihydrogen inorganic compounds suggested possible N₂ reduction mechanisms that involved hydrides.^{83,84,104–108} Such mechanistic speculations included that N₂ binds to Mo-nitrogenase by displacing H₂ through a Mo-dihydride or trihydride intermediate,^83,104 that an H₂ complex bound to Mo might play an important role for N₂ binding,¹⁰⁵ and that nitrogenase evolves H₂ through the reductive elimination of a dihydride species bound to either Mo or Fe, favoring the binding and reduction of N₂.¹⁰⁶ Such multiple contradictory models, although insightful and provocative, were neither experimentally established nor excluded at the time they were suggested.

Despite the connection between N₂ reduction and H₂ release implied by the kinetic scheme of Figure 3, and the speculations it generated, in fact it had been neither confirmed nor universally accepted^20,84 that there is an obligatory, mechanistic requirement for formation of one H₂ per N₂ reduced to two NH₃, with the apparent “waste” of 2[e⁻/H⁺] delivered at the cost of 4 ATP hydrolyzed. Beyond that, none of the intermediate states proposed in LT scheme (Figure 3) had been trapped for characterization: the “boxes” connected by the kinetic scheme were empty. Key breakthroughs came after the determination of the structure of the active site FeMo-cofactor^55,56 and the application of a combined strategy using genetic, biochemical, and biophysical methods characterize nitrogenase intermediates trapped during turnover with different substrates as previously reviewed^{21,23,33,36,82,110} and further detailed in the following sections. As a result of these recent studies, seven of the ten states in the kinetic cycle have now been trapped for study (boxed in red in Figure 3).^21,23,33,82

An important step in characterizing catalytic intermediates was the trapping and characterization of an organometallic intermediate allyl alcohol bound as a ferracycle to a single Fe of FeMo-co during turnover with the substrate propargyl alcohol (Figure 4).¹⁰⁹ This was followed by characterization of an electron-accumulation intermediate (FeMo-co spin S = 1/2) trapped during turnover of MoFe protein having the α−70^Val→Ile substitution,¹¹¹ which apparently prevents access by all substrates to the active site, other than protons.¹¹² Because electron delivery cannot be experimentally synchronized during turnover, the number of electrons accumulated by this intermediate (n) was not evident. To determine this, the system was frozen to 77 K to prevent further electron delivery by reduced Fe protein, and then, it was warmed while frozen (cryoannealing) until relaxation toward E₀ is enabled through a process involving 2-electron steps in which an H₂ is released (see Figure 3). This cryoannealing electron-counting protocol showed this state relaxes in two steps with loss of H₂ in each step, first to E₂(2H) and thence to E₀.¹¹³ As shown in Figure 3, this identifies the intermediate as the key E₄(4H) “Janus” state (defined below),²¹ which has accumulated four of the eight reducing equivalents required by eq 1. ^1,2H/⁹⁵ Mo ENDOR spectroscopy further showed that these equivalents are stored as two [Fe-H-Fe] bridging hydrides,^111,114 not as a previously imagined Mo-H species.^83,104 The E₄(4H) state of the native/wild-type (WT) enzyme itself¹¹⁵ exhibits identical EPR/ENDOR and photophysical characteristics to this state when trapped in the α−70^Val→Ile/α−195^His→Gln altered protein.¹¹⁶ As the WT intermediate accumulates in lower abundance, it has proven to be convenient to have these faithfully equivalent versions of the WT state.^115,116

Figure 4.

Propargyl alcohol reduction intermediate with hyperfine couplings indicated. Reproduced with permission from ref 109. Copyright 2004 American Chemical Society.

3.1. Reductive Elimination/Oxidative Addition (re/oa) Mechanism: As Proposed

The E₄(4H) intermediate represents a critical transition point in the N₂ reduction pathway, Figure 3; as both the culmination of the catalytic phase of electron/proton accumulation by the enzyme and the initial step in the N₂ reduction phase, it is poised to “fall back” to the E₀ resting state by successive release of two H₂,¹¹³ but equally poised to eliminate H₂ and proceed to reduce N₂ to two NH₃ through the accumulation of four more equivalents, hence the name, Janus, for the Roman god of transitions.²¹ The trapping of this intermediate for study opened, for the first time, the possibility of directly testing for/characterizing the proposed equilibrium between two E₄-level states that interconvert through N₂ binding/release coupled to H₂ release/binding without additional [e⁻/H⁺] input, as visualized in Figure 3.

The experimental determination that E₄(4H) stored four reducing equivalents as bridging hydrides^111,114 offered explanations of numerous features of nitrogenase mechanism that had defied explanation for decades,^20,106 while forging a connection between nitrogenase catalysis and the organometallic chemistry of metal hydrides.^{106,108,117–119} See the earlier reviews for an account of early speculations about a possible role for hydrides in nitrogen fixation.^20,84

The connection thus established to the chemistry of metal hydrides offered a potential explanation for and justification of the previously mysterious stoichiometry incorporated in eq 1. This explanation was proposed in a pair of “white papers” that showed how the obligatory H₂ formation during nitrogenase N₂ reduction might explain key experimental observations related to mechanism.^21,23 Once it was recognized that E₄(4H) contains two bridging hydrides, then the chemistry of metal-dihydride complexes^117–119 identifies the proposed [E₄(4H) + N₂] ↔ [E₄(2N2H) + H₂] equilibrium (Figure 3) as representing the reductive elimination (re) of H₂ and its reverse, the oxidative addition (oa) of H₂: a mechanistically coupled re/oa equilibrium Figure 5. The re and release of H₂ drives the most difficult (first) step of N₂ cleavage: reduction of the N₂ triple bond to a diazene level (2N2H) intermediate through delivery to the substrate of the remaining two reducing equivalents.²¹ As mentioned above and discussed in detail below, this proposal also offered an explanation of the key constraints on a mechanism for nitrogen fixation that had been developed over decades of study.²⁰

Figure 5.

Schematic of re/oa equilibrium. In the indicated equilibrium, the binding and activation of N₂ is mechanistically coupled to the re of H₂, as described in the text. Adapted with permission from ref 116. Copyright 2016 American Chemical Society.

3.2. Demonstration that Nitrogenase Functions Through the re/oa Equilibrium

The process of defining the mechanistic core of N₂ activation, the re/oa equilibrium, as visualized in Figure 5, involved multiple steps over a number of years. It included the demonstration that high-spin (S = 3/2) intermediates previously trapped during turnover under argon of wild type enzyme are E₂(2H) isomers,^120–123 and was completed by a capstone study of native nitrogenase that trapped and monitored both E₄(4H) and E₄(2N2H) as a function of turnover conditions.^{22,115,116,124,125} In so doing, it completed the process of establishing that native nitrogenase is activated by the re/oa equilibrium, which mechanistically requires the loss of one H₂ per N₂ and results in the attendant 8e⁻ stoichiometry shown in eq 1.^19,23

Having identified the EPR spectra of all of the states involved in the relaxation of the E₄(4H) and E₄(2N2H) (Scheme 1), it was possible to monitor the kinetics of the relaxation of WT E₄(2N2H) to the resting-state E₀.^115,124 The progress curves for the relaxation of E₄(2N2H) and kinetically linked intermediates during cryoannealing of WT enzyme at −50 °C presented in Figure 6 are well described by the curves calculated from fits to the sequential kinetic scheme involving the re/oa equilibrium of Scheme 1.^115,124 Not only does E₄(2N2H) relax to E₄(4H), but the amount of trapped E₄(2N2H) increases with increasing [N₂], while the amount of E₄(4H) decreases (Figure 7).^115,124

Scheme 1. Kinetic Scheme for the Decay of Freeze-Trapped E₄(2N2H) Derived from Figure 3^a.

Reproduced with permission from ref 124. Copyright 2015 American Chemical Society. ^ak_r and k_b are the second-order rate constants for re and its reverse; k_d and k_d′ are the rate constants for the irreversible decay of E₄(4H) and E₂(2H), respectively.

Figure 6.

Time courses of four EPR detected states during −50 °C cryoannealing of WT low P(N₂) ~ 0.05 atm turnover in H₂O. The data colors correspond to those in the kinetic scheme (top) and the lines correspond to fits to that scheme. Stretched exponential exp(−(t/τ)^m) parameters of the first two fast steps are τ₁ = 43 min, m₁ = 0.79 and τ₂ = 6 min, m₂ = 0.8, and the third slow step fitted as exponential with τ₃ = 330 min. Reproduced with permission from 115. Copyright 2016 American Chemical Society.

Figure 7.

X-band EPR spectra of WT nitrogenase turnover samples trapped under 1 atm of N₂ (with stirring to facilitate transfer of H₂ formed during turnover into the headspace¹²⁴), and under low P(N₂) in H₂O buffer (without stirring) shown in comparison with spectrum of E₄(4H) state trapped during turnover of α−70^Val→Ile MoFe protein of the same concentration. Reproduced with permission from ref 115. Copyright 2016 American Chemical Society.

Correspondingly, the rate of relaxation increases with increasing [H₂], and decreases with increasing [N₂].^115,124 This kinetic coupling of the loss of E₄(2N2H) to the formation of E₄(4H) through oa of H₂, directly establishes the operation of the re/oa equilibrium and its kinetic reversibility.

The observation of both E₄(4H) and E₄(2N2H) in samples freeze-quenched during N₂ fixation allowed measurement of the equilibrium constant for re/oa, K_re = k_r/k_b (Figure 5; Scheme 1).¹¹⁵ This was done by estimating the relative amounts of E₄(4H) and E₄(2N2H) freeze-trapped under experimentally varied partial pressures of N₂ and H₂, yielding the re/oa equilibrium constant, K_re and the free energy for the equilibrium, eqs 4 and 5

$$K_{re}=\frac{{\text{E}}_4\left(2\text{N}2\text{H}\right)}{{\text{E}}_4\left(4\text{H}\right)}\cdot \frac{P\left({\text{H}}_2\right)}{P\left({\text{N}}_2\right)}$$ (4)

$${\Delta }_{re}{G}^{\circ }=-RT\text{ ln}\left(K_{re}\right)~-2\text{kcal}/\text{mol}$$ (5)

Thus, this first step in the reduction of N₂ by nitrogenase is essentially thermoneutral, Δ_reG⁰ ~ −2 kcal/mol, whereas direct hydrogenation of gas-phase N₂ is endergonic by more than 50 kcal/mol.¹¹⁵

3.2.1. re/oa Mechanism Generates the Key Constraints Revealed by Experiment.

Decades of study had revealed several puzzling features of nitrogen fixation by nitrogenase whose mechanistic origins were not understood, but ultimately served as key constraints on any proposed mechanism, Chart 1.^20,21,23 A series of experiments has shown that all of these constraints are generated by the functioning of the re/oa mechanism.

Chart 1.

Key Constraints on the Nitrogenase Mechanism²⁰

These constraints arise from previously confusing results from turnover in the presence of D₂ or T₂. It was found that D₂ and T₂ only react with nitrogenase during turnover under an N₂/D₂ or N₂/T₂ atmosphere^90,98 at the E₄(2N2H) reduction level (c); in this reaction D₂ is stoichiometrically reduced to two HD (with H₂O buffer) (a);^88,89,98 corresponding turnover with T₂ does not lead to exchange of T⁺ into H₂O solvent (b).⁹⁰ It has now been shown that all these D₂/T₂ constraints arise through the reverse of the re process, Figure 5, namely, the oxidative addition (oa) of H₂ with loss of N₂.^21,23 According to the mechanism, during turnover under D₂/N₂, reaction of the E₄(2N2H) intermediate with D₂ generates dideutero-E₄ with two [Fe-D-Fe] bridging deuterides that do not exchange with solvent,²² and the same is true for T₂. These E₄(4H) isotopologues, which are denoted E₄ (2L⁻;2H⁺), L = D or T, would relax through E₂(L⁻;H⁺) to E₀ with successive loss of two HL.^21,113

The proposed formation of the E₄(2D⁻;2H⁺) and E₂(D⁻;H⁺) states through the thermodynamically allowed reverse of re, the oa of D₂ with accompanying release of N₂, in fact was the first successful test of the re/oa mechanism. When these states were formed during turnover under N₂/D₂, the nonphysiological substrate acetylene (C₂H₂) intercepted them to generate deuterated ethylenes (C₂H₃D and C₂H₂D₂).²² More recently the E₄(2D⁻;2H⁺) and E₄(2H⁻;2D⁺) states were observed and characterized by EPR spectroscopy and by studies of isotope effects on the photophysical properties of these isotopologues.¹¹⁵

3.2.2. There Are No Reductive Activation Steps Prior to Catalysis.

In recent years, mechanistic proposals have posited that multiple [e⁻/H⁺] must be delivered to the true resting-state FeMo-co to activate it to a state that acts as the E₀ in an 8[e⁻/H⁺] catalytic cycle equivalent to the scheme of Figure 3.¹²⁶ First, of course, no such preactivation process was observed in the many presteady state measurements carried out on as-isolated nitrogenase over decades.^20,31 But most importantly the measurements of the relaxation of the odd-electron S = 1/2 E₄(4H) with release of 2H₂ (Figure 6)¹¹³ unambiguously show that it relaxes directly to the same EPR-active odd-electron E₀ (S = 3/2) resting state that has been isolated and characterized crystallographically, not to a different state, which at least in one model, would not even be odd-electron.¹²⁶ And there is no doubt that E₄(4H) indeed relaxes to the crystallographically characterized E₀ state. The identity of the crystalline E₀ and the putative E₀ formed by E₄(4H) relaxation is established by the identity of the EPR spectra of the solution E₀ formed by relaxation and that of the crystalline E₀,^57,127 both of which precisely match the long-established EPR spectrum of as-isolated Mo-nitrogenase in solution.

3.3. Exploring the re/oa Mechanism

Upon establishing that nitrogen fixation by nitrogenase operates through the re/oa mechanism with its 8[e⁻/H⁺] stoichiometry, further experiments and computation have been used to explore the microscopic details.

3.3.1. Reductive Elimination Involves an H₂-Bound Intermediate.

The demonstration that the E₄(4H) Janus intermediate has accumulated four reducing equivalents as two [Fe-H-Fe] bridging hydrides, and the resulting link thus forged between nitrogenase catalysis and the organometallic chemistry of metal hydrides, inspired examination of the behavior of E₄(4H) under photolysis as a test of this linkage and as a means of obtaining atomic-level details of the re activation process. The photolysis of transition metal dihydride complexes typically results in the release of H₂,^{108,128–135} and it was found that H₂ release in fact occurred during in situ 450 nm photolysis of E₄(4H) in an EPR cavity at temperatures below 20 K.^116,125 These studies, as supported by density functional theory (DFT) calculations,¹³⁶ showed that an H₂ complex on the ground adiabatic potential energy surface E₄(H₂;2H) becomes populated during catalysis, but this state cannot lose H₂ because the resultant, denoted [E₄(2H)* + H₂], is at inaccessibly high energy.¹²⁵ However, N₂ can bind and displace the H₂ of E₄(H₂;2H) to complete the conversion of [E₄(4H) + N₂] into [E₄(2N2H) + H₂], as represented in Figure 8.¹³⁶

Figure 8.

Cartoon version of suggested catalytic pathway for re/oa activation of FeMo-co for N₂ reduction. Reproduced with permission from ref 125. Copyright 2017 American Chemical Society.

3.3.2. E₂(2H) States 1b and 1c Are Hydride Isomers.

As a further use of hydride photolysis to explore nitrogenase function, it was shown that E₂(2H)/1b stores the two reducing equivalents as a hydride, almost surely with the [Fe-H-Fe] bridging mode of binding, and that 1c is a hydride isomer of E₂(2H)/1b, either with the hydride bridging a different pair of Fe or having converted to a terminal hydride, as visualized in Figure 9.¹²³ Relatively modest differences in the occupancy ratio of the two E₂(2H) hydride conformers during turnover, 1b/1c ~ (2–3)/1, suggested that the two are able to readily interconvert during fluid-solution turnover.¹²³

Figure 9.

Schematic representation of photolysis conversion of [Fe-H-Fe] hydride bridge of E₂(2H)/1b into isomer E₂(2H)/1c with either a hydride bridge between a different pair of Fe ions or an Fe-H terminal hydride. Adapted with permission from ref 123. Copyright 2018 American Chemical Society

3.3.3. Density Functional Studies of the re/oa Equilibrium.

The structure and energetics of possible nitrogenase intermediates were recently explored by DFT calculations on a broad range of structural models for the active site and a variety of DFT exchange and correlation functionals.^136–139 These computations¹³⁶ showed that it is necessary to include all residues (and water molecules) interacting directly with FeMo-co via specific H-bonds interactions, nonspecific local electrostatic interactions, and steric confinement to prevent spurious disruption of FeMo-co^{126,140–142} upon accumulation of 4[e⁻/H⁺]. These calculations indicated an important role of sulfide hemilability in the overall conversion of E₀ to a diazene-level intermediate, as suggested by X-ray crystallographic studies.^{58,59,143,144} Perhaps most importantly, they explained how the enzyme mechanistically couples exothermic H₂ formation to endothermic cleavage of the N≡N triple bond in the nearly thermoneutral re/oa equilibrium discovered experimentally (Figure 10), while preventing the futile generation of two H₂ without N₂ reduction: hydride re generates an H₂ complex, again in agreement with experiment, and H₂ is only lost when displaced by N₂, to form an end-on N₂ complex, which then proceeds to a diazene-level intermediate.¹³⁶

Figure 10.

Energetics of formation of E₄(2N2H) from the enzymatic reactants and its decomposition into two individual two-electron, two-proton processes, each involving the hydrolysis of 2ATP per electron: the formation of E₄(2N2H) and of H₂, which are coupled through displaces and releases H₂ formed by re from E₄(4H). Adapted with permission from 136. Copyright 2018 National Academy of Sciences.

3.3.4. High-Resolution ENDOR Spectroscopy Coupled with Quantum Chemical Calculations Reveals the Structure of Janus Intermediate E₄(4H).

Because of the central role of E₄(4H) in the re/oa mechanism, it was essential to characterize its structure. The original ¹H ENDOR study of E₄(4H),¹¹¹ and subsequent ⁹⁵Mo and ⁵⁷Fe ENDOR study^114,145 demonstrated that FeMo-co in E₄(4H) contains two [Fe-H-Fe] bridging hydrides but were unable to provide a satisfactory picture of the relative spatial relationships of the two hydrides. The nature of E₄(4H) has been the subject of numerous recent computational studies based on DFT.^136–139 The lowest-free-energy E₄ isomers obtained from the DFT studies reported so far have two of the four hydrogens located on μ₂-S atoms, while the other two hydrogens are either present as hydrides in bridging positions between two Fe atoms, or in mixed bridging/terminal position on the Fe_2,3,6,7 face of FeMo-co,^{136,137,146,147} or as a dihydrogen (η²-H₂) adduct bound to either Fe₂^136,137,147 or Fe₆.¹⁴⁷

As we pointed out, limitations in DFT-based schemes—the only ones computationally feasible for a systematic study of complex systems, such as FeMo-co—make it difficult to determine the ground-state structure of E₄(4H) based on DFT-derived relative energies alone. To emphasize this issue, careful computational analyses carried out by the research groups of Ryde^{137,146,148,149} and Bjornsson^139,147 have clearly shown that the exchange and correlation functional adopted in the DFT calculations largely affect the ordering of possible E₄ isomers, which differ on the location of the hydrides. This uncertainty is further aggravated by intrinsic deficiency of broken symmetry DFT calculations which are simply unable to capture the multideterminant character of the electronic structure of FeMo-co. A detailed discussion of all of these issues has been presented by Cao and Ryde.¹⁴⁶

To determine how the Janus intermediate binds its two hydrides, exhaustive, high-resolution CW-stochastic ¹H-ENDOR experiments were performed using improved ENDOR methodologies and updated equipment.¹⁵⁰ Figure 11 illustrates the remarkable spectroscopic resolution and strong orientation selection that enabled the previously unattainable determination of absolute hyperfine-interaction signs. The important results for structural assessments were the findings that (i) the dipolar–electron–nuclear hyperfine coupling tensors for the two hydrides were coaxial, (ii) each had “rhombic symmetry” (two components with equal and opposite signs; one null component), and (iii) the signed tensor components for the two hydrides were permuted relative to each other.

Figure 11.

35 GHz ¹H stochastic-field modulation detected (stochastic CW) ENDOR spectra of E₄(4H) in 70^Val→Ile/α−195^His→Gln MoFe protein, acquired at g = 1.991 and 1.964 (black) and summed simulations (red). The signal with ν₀(¹H) ± ~3 MHz, with both positive and negative features, represents transient responses from weakly coupled, more distant protons. Reproduced with permission ref 150. Copyright 2019 American Chemical Society.

These findings provided critical constraints on the structure of E₄(4H), but did not by themselves implicate a particular structure, and so experiment was coupled to the DFT computations. Just as the DFT computations could not discriminate among candidates for the E₄(4H) ground state, neither could they discriminate through computations of the dipolar hyperfine couplings. However, the DFT structures are robust and so the ENDOR measurements were coupled to DFT structural models through an analytical point-dipole Hamiltonian for the hydride electron–nuclear dipolar coupling to its “anchoring” Fe ions, an approach that overcomes limitations inherent in both experimental interpretation and computational accuracy. The simple point-dipole approximation was not expected to provide quantitatively accurate results for the magnitudes of the hyperfine tensor components because of the delocalized nature of the electron density in the FeMo-co.¹⁴⁷ Instead, for each of the DFT-derived candidate structures for E₄(4H) ground state the dipole approach was simply used to assess the relative orientation of the dipolar interaction tensors for the two hydrides of a structural model when each had a rhombic tensor, and this was compared to the experimentally determined relative tensor orientation.^146,147,150

This comparison alone enabled discrimination among possible E₄(4H) isomers.^23,111 The result: the freeze-trapped, lowest-energy Janus intermediate structure is E₄(4H)^(a), Figure 12. As the calculations indicated the presence of a number of isomers close in energy,^{137,139,146,147,150} it was envisioned that at ambient temperatures the E₄(4H) hydrides become fluxional, with a dynamic population of multiple isomers.

Figure 12.

Cartoon of the 2,3,6,7 face of the ground state isomer of E₄(4H), denoted E₄(4H)^(a) as found by BS-DFT computation. Reproduced with permission from ref 150. Copyright 2019 American Chemical Society.

As additional DFT computations are carried out, suggesting variations on the E₄(4H) structure, the results of this ENDOR study, in conjunction with the analytical point-dipole Hamiltonian, will continue to provide the benchmark against which those structures can be tested.

3.3.5. E₁ Forms by Reduction of Fe, Not Mo.

The E₁(H) state as prepared both by low-flux turnover and by radiolytic cryoreduction were investigated using a combination of Mo Kα-HERFD (high energy resolution fluorescence detection) and Fe K-edge PFY (partial-fluorescence yield) X-ray absorption spectroscopy techniques.¹⁵¹ The results demonstrate that this state is formed by an Fe-centered reduction and that Mo does not change redox state, a further indication that Mo is only indirectly involved in substrate reduction, which occurs at Fe. This conclusion is further supported by a recent Mo and Fe K-edge EXAFS (extended X-ray absorption fine structure) characterization of E₀ and E₁ states in combination with QM/MM calculations that supported protonation of a belt sulfide (S2B or S5A) in E₁.¹⁵²

3.4. Kinetic Description of the Competing Reactions at E₄(4H)

As can be seen in Figure 13, the reactions of E₄(4H) can be viewed as a competition: the forward direction of the re/oa equilibrium, involving binding of N₂ and displacement of H₂, competes with the relaxation of E₄(4H) through hydride protonolysis (HP), which releases H₂ without N₂ binding.^153,154 The competition between these two reactions is defined by the ratio of their rate constants k_re for the forward reaction and k_HP for the relaxation reaction and modulated by the ratio of the partial pressures of N₂ and H₂. A kinetic model was developed that yields, under high electron flux conditions (high Fe protein to MoFe protein molar ratio), the values of these two rate constants from the fitting of the ratio, H₂ formed divided by N₂ reduced, as a function of the partial pressure of N₂. As can be seen in Figure 14¹⁵⁴ and from measurements by many groups, at 1 atm of N₂, Mo-nitrogenase shows H₂/N₂ < 2 and the ratio asymptotically approaches unity with increasing N₂ partial pressure. The fit to the data of Figure 14 yielded k_re/k_HP = 5.1 atm⁻¹. This considerable preference for the forward reaction is truly remarkable when considering the expected reactivity of a metal hydride toward protonolysis as observed in many reductive catalysts. Corresponding suppression of the formation of H₂ by an HP reaction in synthetic catalysts for N₂ reduction is one of the grand challenges in catalyst design.

Figure 13.

Reactions of the E₄ state. N₂ binds through re with rate constant k_re to the right. H₂ evolves through HP with rate constant k_HP to the left. Reproduced with permission 153. Copyright 2018 American Chemical Society.

Figure 14.

Ratio of H₂ evolved to N₂ reduced as a function of partial pressure N₂ for all three nitrogenase isozymes. The fits to this data for each enzyme produce the value of its parameter ρ = k_re/k_HP Reproduced with permission from ref 154. Copyright 2019 American Chemical Society.

The two other isozyme of Mo-nitrogenase, the V- and Fe-nitrogenases, share similar architecture to the Mo-enzyme, but as noted above, have different polypeptides that present different active site environments and different active site metal clusters.^43,58 At 1 atm of N₂, it can be seen that the V- and Fe-nitrogenase show much higher ratios of H₂ formed per N₂ reduced (Figure 14). Application of the high-flux kinetic model to the V- and Fe-nitrogenases, revealed reduced values of the rate-constant ratio compared to Mo-nitrogenase, k_re/k_HP= 1.1 and 0.78 atm⁻¹, respectively.¹⁵⁴ Further analysis pinpointed that whereas the values for k_re for the V- and Fe- nitrogenases are as much as 10-fold smaller than k_re for the Mo enzyme, the values of k_HP differ by less than 2-fold. In short, the V- and Fe-nitrogenases are not better at the HP side-reaction, but rather are poorer at the productive re reaction. The source of these differences among the three nitrogenase isozyme remains to be deduced.^153,154

3.4.1. All Three Nitrogenases Employ the re/oa Mechanism.

The differences in rate constants among the three isozymes might suggest a different mechanism for N₂ binding and activation. However, this is not so. The re/oa mechanism involves an equilibrium between the E₄(4H) state and the E₄(2N2H) state. This predicts H₂ inhibition of N₂ reduction and the formation of HD during turnover under mixed D₂ and N₂, both of which were found for all three isozymes. These observations establish that all three nitrogenase isoforms follow the same re/oa mechanism.^38,153,154

3.5. Structure–Function Relationship During Nitrogenase Catalysis

One of the persistent challenges in nitrogenase research is to understand how protein and cluster dynamics might participate in the mechanism.^{32,33,35–37,68,75,156–168} Many structural and spectroscopic studies have shown that the resting state of nitrogenase is largely unreactive, suggesting that motion of the surrounding protein or changes in ligation of FeMo-co might be essential in catalysis.^32,33,82,158 A number of calculations have suggested changes in FeMo-co that could be occurring during catalysis, including breaking of one or more the Fe–C bonds^{126,140,141,169–171} or breaking one or more Fe–S bonds.^{136,172–177} Recent structural studies have observed significant changes in the FeMo-co and FeV-co structures that might point to dynamics relevant to catalysis.^{58,59,143,144,178} In one such study, it was observed that MoFe protein under turnover with CO shows the substitution of one of the belt sulfide ligand S2B by a bridging CO ligand¹⁴³ (see section 5.2 for more details). These studies point to lability in one or more of the Fe–S bonds during turnover.^178,179 This lability was seen in the computational study of the re/oa equilibrium described in sections 3.3.3 and 3.3.4.¹³⁶ Recently, a structure of the V-nitrogenase by Einsle et al. showed FeV-cofactor with two belt sulfide ligands substituted by a proposed carbonate and a light atom (N(H) or O(H), Figure 15).^59,155 Even though the identity of the light atom ligand is under debate, the plausible implication and relevance to the N₂ reduction mechanism by nitrogenase have initiated vigorous discussion.^156,180,181 The recent progress in spectroscopic studies on nitrogenase intermediates¹⁶² and the X-ray structures of nitrogenase from turnover mixtures promises further progress in understanding the dynamics of the metalloclusters, especially the active site FeM-cofactor during substrate/inhibitor interaction.

Figure 15.

FeV-cofactor structure proposed by X-ray structure (X = NH)⁵⁹ and theoretical calculations (X = OH).¹⁵⁵

4. NONPHYSIOLOGICAL N-BASED SUBSTRATES

4.1. Range of N-Based Substrates

An important property of nitrogenases is the ability to bind and reduce a number of nonphysiological substrates.^20,39,40 A common, but not universal, property of nitrogenase substrates is that they are small and contain double or triple bonds. The study of how these alternative substrates interact with nitrogenase has provided important mechanistic insights.^{20,23,33,41,82} In this section, recent work on a number of alternative substrates that contain N is reviewed.

4.2. N₂H_x Substrates

As shown in the re/oa kinetic scheme of Figure 3, following the formation of E₄(2N2H), the catalytic cycle traverses four states that contain partially reduced species of N₂.^21,23 Researchers have long considered two competing proposals for this “second half” of the kinetic scheme, which invoke distinctly different intermediates, Figure 16. In the “distal” (D) pathway, utilized by N₂-fixing inorganic Mo complexes,^83,183–185 the distal N of Fe-bound N₂ is hydrogenated in three steps. The first NH₃ is then liberated with the addition of the fifth [e⁻/H⁺] overall, to generate the nitrido species, E₅. This nitrido-N is then hydrogenated three times to yield the second NH₃. In the “alternating” (A) pathway that has been suggested to apply to reaction at Fe of FeMo-co,^175,186 the two N’s instead are hydrogenated alternately, with a four-electron and proton reduced species equivalent to hydrazine (N₂H₄) formed as the E₆ intermediate. In this case, the first NH₃ is only liberated during the next hydrogenation step, the formation of E₇ (Figure 16).^33,82 Peters et al. has proposed what might be called a hybrid-A pathway,¹⁸² in which E₄ on the D pathway is converted to E₅ on the A pathway. Consequently, as with the limiting A pathway, the first NH₃ also is released with formation of E₇, as shown.

Figure 16.

Proposed N₂ reduction pathways with M representing the active site FeMo-cofactor. A distal reduction pathway (D) and an alternating reduction pathway (A) are displayed with plausible key intermediates. The hybrid-A pathway is indicated by a dashed arrow.¹⁸²

Previous studies demonstrated that the diazene analogs methyldiazene (CH₃-N=NH), dimethyldiazene (CH₃-N=N-CH₃) and diazirine (CH₂N₂) are substrates for nitrogenase, all being reduced to NH₃, and CH_x species.^187,188 The early studies on these substrates have been reviewed.⁹³ In 2007, the study of diazene (HN=NH) as a substrate for Mo-dependent nitrogenase was expanded through the use of in situ generation from azodiformate.¹⁸⁹ The 4 [e⁻/H⁺] reduction of diazene produced two equivalents of NH₃. The reduction of diazene was found to be inhibited by H₂, which is also an inhibitor of N₂ reduction.¹⁸⁹ As can now be understood, in both cases the inhibition is a consequence of the re/oa mechanism as described in section 3. This result implies that diazene must enter the reaction pathway by forming the same diazene-level intermediate, E₄(2N2H), as forms when N₂ reacts with E₄(4H).²³ The addition of H₂ then reverses the re/oa equilibrium, with the E₄(2N2H) undergoing oa with loss of N₂ to generate E₄(4H). The E₄(2N2H) could form in one of two ways: direct reaction with E₀ or, as proposed in 2012,¹⁹⁰ reaction with E₂(2H) with loss of H₂ through hydride protonation.

Recently it has also been shown that the 2 [e⁻/H⁺] reduction of hydrazine to produce two equivalents of NH₃¹⁹¹ is efficiently achieved by remodeling of the α−70^Val residue around FeMo-cofactor of Mo-nitrogenase with a smaller side-chain amino acid (Ala).¹⁹² In contrast to the H₂ inhibition of diazene and N₂ reduction, hydrazine reduction was not inhibited by H₂. This again is consistent with the scheme of Figure 3. Once the delivery of [e⁻/H⁺] to E₄(2N2H) has generated E_5, the enzyme has passed the re/oa equilibrium stage and is no longer susceptible to inhibition by H₂, while hydrazine appears on the A N₂ reduction pathway at E₆.¹⁹⁰

Two common intermediates have been freeze-trapped using MoFe protein variants with both diazene substrates (CH₃-N=NH and HN=NH) and hydrazine N₂H₄ (Figure 17):^{188,189,192,194} a broad, low-field resonance H (non-Kramers state, S ≥ 2) EPR-detectable in Q-band at 2 K; a Kramers state, S = 1/2, with g = [2.09, 2.01, 1.98]), denoted, I.¹⁹⁰ These intermediates have been fully characterized by X/Q-band EPR and ¹⁵N, ^1,2H-ENDOR, HYSCORE, and ESEEM spectroscopic studies. Their enzymatic and chemical assignments are E₇(NH₂) = H with NH₂⁻ bound to FeMo-cofactor and E₈(NH₃) = I, with a bound NH₃.¹⁹⁰

Figure 17.

Q-band CW EPR spectrum of α−70^Val→Ala/α−195^His→Gln MoFe protein in the resting state and trapped during turnover with ¹⁴N₂H₄ as substrate. Kramers intermediate I and non-Kramers intermediate, H, are noted in the turnover spectrum. Reproduced with permission from ref 190. Copyright 2012 Authors. Published by PNAS.

4.3. NO_x Substrates

Nitrite (NO₂⁻), nitrous oxide (N₂O), and hydroxylamine (NH₂OH) all are reduced by nitrogenase.²⁰ The recent characterization of nitrite as a nitrogenase substrate established that six-electron reduction of NO₂⁻ to NH₃ by nitrogenase proceeds without the obligatory H₂ evolution that is required for the reduction of N₂.¹⁹³ The kinetic observations and mechanistic implications of the first three of these substrates have been summarized in detail in an earlier review.²⁰ Hydroxylamine was recently reported as a nitrogenase substrate, being reduced by two electrons to NH₃ and H₂O.¹⁹³ The reduction of each of these substrates is not inhibited by H₂, as in the case in the reduction of hydrazine. As with hydrazine, these results are as expected for reactions that involve intermediates that correspond to E_n, n > 4 (see above).

Freeze-trapping α−70^Val→Ala/α−195^His→Gln MoFe protein under turnover conditions with NO₂⁻ or NH₂OH as substrate resulted in the same two EPR-active species (S = 1/2 and S ≥ 2) trapped with diazenes or hydrazine as substrate (Figure 18). As confirmed by ¹H and ¹⁵N pulsed ENDOR study, the S = 1/2 intermediates of NO₂⁻ or NH₂OH are E₈(NH₃) = I; ^14/15N ESEEM studies confirmed the identity of the non-Kramers intermediates as E₇(NH₂) = H (S ≥ 2) for NO₂⁻ or NH₂OH.¹⁹³

Figure 18.

X-band EPR spectra of α−70^Ala/α−195^Gln MoFe protein turnover samples prepared with N₂H₄ (black), NO₂⁻ (red), and NH₂OH (green) substrates. Reproduced with permission 193. Copyright 2014 American Chemical Society.

4.4. Mechanisms and Insights

The study of these N-based alternative substrates has provided evidence regarding the nature of the reaction pathway for N₂ reduction: D or A (Figure 16). An alternating pathway for N₂ hydrogenation is supported by the observation that (i) hydrazine is a product of N₂ reduction by V-nitrogenase,¹⁹⁵ which also functions through the re/oa mechanism,^153,154 and (ii) hydrazine was detected as a product by acid or base quenching of Mo-nitrogenase during N₂ reduction.^86,87 Of course, as seen in Figure 16, in the D pathway the N-N bond is cleaved before hydrazine can be formed and so could not be formed during catalysis, thus indicating that the A pathway is followed by nitrogenase.

The kinetic and spectroscopic observations of nitrite and hydroxylamine reduction by nitrogenase led to the proposal of a [6e⁻/7H⁺] reduction mechanism of nitrite by nitrogenase (eq 6) as shown in the right panel of Figure 19.¹⁹³ In this mechanism, NO₂⁻ and one proton bind to FeMo-cofactor at the E₂ state. After one water molecule was released, the resulting M-[NO⁺] species reacts with the hydride on the same cofactor to avoid the formation of an M-[NO] “thermodynamic sink”.¹⁹⁶ The formation of an M-HNO species is similar to the reduction of heme-bound NO to bound HNO through hydride transfer from NADH by P450 NO reductase.¹⁹⁷ The M-HNO intermediate is then converted to further reduced species through a pathway analogous to the nitrite reduction by cytochrome c nitrite reductase (ccNIR),^198–201 including intermediates H (M-NH₂) and I (M-NH₃).¹⁹⁰

$${\text{NO}}_2^{-}+6{\text{e}}^{-}+7{\text{H}}^{+}\to {\text{NH}}_3+2{\text{H}}_2\text{O}$$ (6)

Figure 19.

Comparison of the proposed reduction pathways of N₂ (left) and nitrite (right) by nitrogenase (M represents FeMo-cofactor). An intermediate, labeled E_n on N₂ pathway and E_m′ on nitrite pathway, has accumulated n or m [e⁻/H⁺]. (Boxed Region) Convergence of pathways for nitrite and N₂ reduction by nitrogenase, as discussed in the text. Within this region, boxed reactions of E₁ show the most direct routes by which N₂H₄ and NH₂OH join their respective pathways. Reproduced with permission from ref 193. Copyright 2014 American Chemical Society.

Even though the binding and reduction mechanism of N₂ and NO₂⁻ are totally different, the generation of common intermediates H and I from both the di-N substrates (N₂H₂ and N₂H₄) and mono-N substrates (NO₂⁻ and NH₂OH) further confirmed the previous assignments of intermediate H to M-NH₂ and I to M-NH_3, and their assignment to E₇ and E₈, respectively.^190,193 These results are understood as arising from a convergence of the reduction pathways of N₂ and NO₂⁻ in the late stage of the corresponding catalytic cycles as highlighted in the red box in Figure 19.¹⁹³

Hydrazine and hydroxylamine were proposed to bind to E₁ and undergo bond cleavage, forming the intermediate H, and NH₃ and H₂O, respectively.^190,193 These results are consistent with the proposed alternating N₂ reduction pathway by nitrogenase in which the first NH₃ is released after delivery of the seventh [e⁻/H⁺] to the N₂H₄-bound E₆ state to form intermediate H. The intermediate H accepts one last pair of [e⁻/H⁺] to form intermediate I, which regenerates the E₀ state to finish one catalytic cycle by releasing the second NH₃.^21,23,190

5. C-BASED SUBSTRATES

In addition to the nonphysiological N-based substrates, nitrogenase interacts with a broad range of carbon (C)-based substrates/inhibitors. The utilization of these C-based substrates/inhibitors has greatly expanded the ability to probe nitrogenase mechanism, and their reduction has been reviewed.^20,39,40,203 In this section, we will focus on the progress in studying C-based substrates/inhibitors of nitrogenase since 2013,⁴⁰ referring to earlier work when necessary for clarity.

5.1. Alkyne Substrates

Alkynes were widely used as substrates and inhibitor in probing the substrate-binding sites and reduction mechanism of nitrogenase over the last half-century.^20,39,40,179 Acetylene was the first alkyne discovered early as an alternative substrate of nitrogenase,²⁰⁴ with pioneering works studying the steady-state EPR of wild type²⁰⁵ and α−195^His→Gln-substituted MoFe protein²⁰⁶ with acetylene as substrate. The X- and Q-band EPR spectra of the freeze-trapped species exhibited three simultaneously generated S = 1/2 signals: a rhombic S_EPR1 signal with g = 2.12, 1.98, 1.95, a rhombic S_EPR2 signal with g = 2.007, 2.000, 1.992, and a minor isotropic S_EPR3 signal with g ≈ 1.97.^206,207 Initial ¹³C and ¹H ENDOR measurements on the S_EPR1 signal suggested that it arises from a substrate complex, with two C₂H₂-derived species bound to the cofactor, one of them being a C₂H₂ bound to the FeMo-cofactor by bridging two Fe ions of the Fe2,3,6,7 face.²⁰⁷ However, results described directly below, which were obtained for an intermediate trapped during turnover of the alkyne propargyl alcohol with the α−70 ^Val→Ala-substituted MoFe protein suggested that this interpretation be reexamined. Subsequent, improved ENDOR measurements on S_EPR1, which included ⁵⁷Fe ENDOR measurements, revealed interactions with three distinct carbon nuclei, again indicating that at least two C₂H₂-derived species are bound to the cofactor, but they suggested that this intermediate in fact contains an ethylene product bound as a ferracycle to a single Fe of a two-electron reduced (E₂) FeMo-cofactor (Figure 20).²⁰²

Figure 20.

Schematic representation of chemical structure of S_EPR1 species at E₂ state, highlighting the ferracycle formed by C₂H₄ binding to Fe₆ as proposed in literature.²⁰² The 4Fe4S face represents the same orientation as in Figure 2.

It was long held that FeMo-cofactor contained within the MoFe protein required the accumulation of one or more electrons before any substrate or inhibitor could interact.^20,209 This conclusion was drawn from the lack of significant perturbations of the resting state (E₀) FeMo-cofactor EPR signal line shape upon the addition of any inhibitors or substrates.²⁰ This situation changed with the report that when the α−96^Arg, located on one FeS face of FeMo-cofactor within the MoFe protein, was substituted by a series of other amino acids, the resulting MoFe protein variants showed perturbations of the FeMo-cofactor EPR signal (S = 3/2, g = 4.26, 3.67, 2.00) upon incubation of the resting state enzyme with the substrate acetylene or the inhibitor cyanide (CN⁻).²⁰⁸ Both acetylene^204,210,211 and cyanide^212–214 were long ago reported as substrates for nitrogenase under turnover conditions, being reduced to ethylene (C₂H₄) and methane (CH₄) and ammonia (NH₃), respectively. Incubation of resting-state α−96^Arg→Leu-substituted MoFe protein with acetylene or cyanide was found to decrease the normal FeMo-cofactor signal with appearance of new EPR signals having g = 4.50 and 3.50 (Figure 21), and g = 4.06, respectively. Subsequent ¹³C ENDOR studies revealed direct, but weak, interactions between acetylene or CN⁻ and the resting-state FeMo-cofactor.²⁰⁸ These observations were the first to show that substrates/inhibitors can interact with the resting state FeMo-cofactor contained within the MoFe protein.

Figure 21.

X-band EPR spectra of α−96^Leu MoFe protein under nonturnover conditions in the absence (top trace) and presence (bottom trace) of acetylene. Reproduced with permission from ref 208. Copyright 2001 American Chemical Society.

Capitalizing on this finding, a recent X-ray crystallographic study of α−96^Arg→Gln-substituted MoFe protein in the presence of acetylene resulted in the first crystal structure with a substrate molecule, acetylene, captured near the active site FeMo-cofactor.²¹⁵ In the 1.70 Å crystal structure, the acetylene is located 4–5 Å away from Fe6 atom, which was supported by DFT calculations (Figure 22).²¹⁵ These results are consistent with the previous findings that nitrogenase binds some substrates at the 4Fe4S face of FeMo-co approached by the side chains of α−70^Val and α−96^Arg.^33,36 An ¹H and ⁵⁷Fe ENDOR study using wild-type and Nif V⁻ MoFe proteins from Klebsiella pneumoniae also revealed that acetylene binding to the resting-state MoFe protein can perturb the FeMo-cofactor environment.²¹⁶

Figure 22.

Comparison of the acetylene location at the FeMo-cofactor. The acetylene location at the FeMo-cofactor is compared between X-ray (A) and DFT optimized structures (B). The interatomic distances are shown with red lines. Reproduced with permission 215. Copyright 2017 Elsevier, Ltd.

The first detailed characterization of alkyne-reduction intermediates, alluded to above, was carried out on states freeze-trapped during the turnover of nitrogenase modified by substitution of the valine at α−70 position in MoFe protein by alanine.^109,217,218 This substitution allows the reduction of longer-chain alkyne substrates,^219,220 such as propargyl alcohol (HC≡CCH₂-OH).²¹⁷ When the α−70^Val→Ala-substituted MoFe protein was frozen in liquid nitrogen during steady-state turnover with propargyl alcohol as substrate, the S = 3/2 EPR spectrum of the FeMo-cofactor was changed to a new S = 1/2 signal.^217,218 Using ^1,2H- and ¹³C-ENDOR spectroscopies, it was demonstrated that this S = 1/2 signal originates from the FeMo-cofactor having the alkene allyl alcohol (H₂C=CHCH₂-OH) reduction product bound as a ferracycle to a single Fe as shown in section 3 (Figure 4).¹⁰⁹

This structural assignment was later supported by the EXAFS and NRVS examination of the same intermediate.¹⁶² A similar S = 1/2 signal was also observed when propargyl alcohol was replaced by propargyl amine (HC≡CCH₂-NH₂) as substrate. Trapping of these two intermediates was shown to be dependent on pH and the presence of histidine at α−195 position.²¹⁸ These results suggested that these intermediates are stabilized, and thereby trapped, by H-bonding interactions between either the −OH group or the −NH₃⁺ group and the imidazole of α−195^His. Theoretical calculations refining the binding mode and site pointed to η² coordination at Fe6 of the FeMo-cofactor, as visualized in Figures 4 and 21.²¹⁸

5.2. CO as Inhibitor and Substrate

CO was described as an inhibitor of wild-type Mo-nitrogenase in early studies, being shown to inhibit the reduction of all substrates except protons to make H₂.^20,93,221 CO inhibition of proton reduction was observed when the first sphere (substitution of homocitrate by citrate as in the NifV⁻ mutant)^222,223 or secondary sphere environment of the FeMo-cofactor^222,224,225 was changed. Similar CO inhibition of proton reduction was also seen for Mo-nitrogenase under high pH conditions²²⁶ and for V-nitrogenase.^227,228 Recently, CO was reported to stimulate the H₂ production catalyzed by three MoFe protein variants, α−277^Cys, α−192^Asp, and α−192^Glu under high-electron flux conditions but not under low-electron flux conditions. These intriguing observations are apparently at odds with the observation that CO displays either no inhibitory or inhibitory effect on proton reduction catalyzed by Mo- and V-nitrogenases, as noted above.²²⁹

Trapping of nitrogenase by freezing during turnover under CO revealed a loss of the EPR spectrum of the resting state FeMo-cofactor and appearance of two major different EPR signals.^{205,230–234} In the presence of low concentrations of CO ([CO]:[FeMo-cofactor] ≤ 1:1), the S = 3/2 spectrum of FeMo-cofactor changed to an S = 1/2 spin state giving rise to a new rhombic EPR signal (denoted lo-CO) with g = 2.09, 1.97, 1.93. At higher concentrations of CO (≥0.5 atm), an axial EPR signal (S = 1/2, called hi-CO) was observed with g = 2.17, 2.06, 2.06.^{205,231–234} In addition to these trapped CO complexes, a small population of species at S = 3/2 spin state, so-called hi(5)-CO, has been observed under high CO concentrations and higher electron flux in wild-type²³³ and variant MoFe proteins.^222,234

These trapped CO complexes of FeMo-cofactor have been well studied by advanced spectroscopic methods. ENDOR studies of lo-CO and hi-CO complexes using different isotopologues of CO and FeMo-cofactor characterized the two CO complexes.^{202,230,235–238} Key conclusions from these studies were that in lo-CO the C of a single CO bridges two Fe ions (end-on mode) and in hi-CO, two CO terminally bind to two Fe ions in the “waist” of the FeMo-cofactor (Figure 23A). The hi(5)-CO complex has been suggested to contain two bridging CO molecules on the FeMo-cofactor.²²² Finally, EPR-monitored photolysis experiments revealed that a terminal CO of hi-CO complex is photolabile, causing the state to reversibly convert to lo-CO. In contrast, both lo-CO and hi(5)-CO complexes are photostable.²³⁹

Figure 23.

Schematic structures of CO complexes as proposed based on (A) EPR/ENDOR,^{202,230,235–238} (B) IR-monitored photolysis,^240–242 and (C) X-ray crystallography of Mo-nitrogenase¹⁴³ and EPR comparison of Mo- and V-nitrogenases.²⁴³ The presence of a bound carbonate in the VFe-co was revealed later by X-ray crystallography.

The E_n level of lo-CO and hi-CO remain a matter of considerable interest. It has been reported that lo-CO and hi-CO can be interconverted by adding CO to the former or pumping CO off of the latter, without redox or catalytic processes, thus demonstrating that they are at the same E_n level.²³³ In addition, lo-CO can be converted to E₀ by extensive pumping, which suggests that all three states are at the same, M^N (E₀), redox level,²³³ an idea consistent with ⁵⁷Fe ENDOR measurements.^145,237 However, CO cannot bind to FeMo-cofactor in the resting state, and can only bind to the cofactor in a state formed during turnover under CO.²³³ One possibility is that lo-CO is indeed an E₂ state, and we have discussed this possibility.²⁰² However, such a state should have the two reducing equivalents stored as a hydride, and no such ¹H ENDOR signal is observed.^150,236 Thus, it appears unlikely that lo-CO is an E₂ state. To explain the combined observations one might imagine that CO binds to FeMo-cofactor in an E₂ state, or in an E₁ state that then receives a second electron, and that the resulting CO-bound E₂ state loses H₂ to form the observed lo-CO, which indeed is at an E₀ redox level. The differing net spins, S = 3/2 for resting state, S = 1/2 for CO-bound states, would then merely reflect the influence of the CO binding. Clearly, a reexamination of the lo-CO state, including ⁹⁵Mo ENDOR and DFT computations, would seem to be in order.

Although EPR and ENDOR give the optimal insights into properties of intermediates with half-integer spin states (S = 1/2, 3/2, 5/2, …), rarely (intermediate H, section 4.2) can it probe intermediates with integer spin (S = 0, 1, 2, …).^190,193 Stopped-flow FTIR studies are equally effective for all spin states.^244–246 Recently, IR-monitored photolysis studies of complexes of wild-type and two MoFe protein variants trapped during turnover with low electron flux under hi-CO conditions displayed a more complex landscape of hi-CO complexes.^240–242 Three different hi-CO complexes (designated as Hi-1, Hi-2 or hi(5)-CO, and Hi-3) were observed with absorptions at different energy levels. All hi-CO complexes have two CO bound to FeMo-cofactor and can reversibly release one CO to generate lo-CO complexes with one terminal, bridging, or protonated CO species. The three hi-CO complexes all have one CO terminally bound to FeMo-cofactor with the other ligand as a terminal, semibridging, or bridging/protonated CO ligand (Figure 23B).^240–242 Further photolysis and DFT study of Hi-3 with C-O stretching frequencies at 1938 and 1911 cm⁻¹ revealed that it has two terminal CO ligands bound to two adjacent Fe atoms in the FeMo-cofactor with an EPR silent S = 0 spin state.^241,242 DFT calculations of the C-O stretching frequencies of hi-CO complexes also suggested that C-O frequencies at 1973 and 1680 cm⁻¹ in IR of Hi-1 species originated from a terminal −CO and a partially reduced −CHO ligands bound to two adjacent Fe sites as well, consistent with the EXAFS and NRVS observations for the wild-type enzyme.²⁴² Recent NRVS studies of nitrogenase under high CO conditions revealed that the FeMo-cofactor undergoes significant structural perturbation upon −CO/−CHO binding.²⁴² The identification of Fe-CO stretching and Fe-C-O bending modes in the range of 470–560 cm⁻¹ in NRVS provided the first confirmation that binding is indeed at Fe sites,^23,110 extending the many studies noted above that showed the heterometal, Mo in FeMo-cofactor, may tune the properties of the cofactor, but is not directly involved in valence changes or substrate reduction.^114,151

As described above, structures of the FeMo-cofactors with CO or CO-derived ligands have been proposed based on the different spectroscopic studies (Figure 23). In 2014, Rees et al.¹⁴³ reported the first X-ray crystal structure of a CO-bound state of the MoFe protein with 1.5 Å resolution, using crystals obtained from a solution, following turnover in the presence of CO and C₂H₂. Surprisingly, the structure showed the bridging CO proposed for lo-CO, but it was found to bind to the FeMo-cofactor by replacing one of the belt sulfide ligands, S2B (Figure 23C). Correspondingly, a recent EPR study of the CO-modified MoFe protein generated under the same conditions that yielded the CO-modified crystals yielded a spectrum that almost exactly matches that reported for lo-CO.²⁴³ The dissociation of a belt sulfide is a central factor in the developing view, noted above, that FeMo-cofactor is not structurally rigid^71,143 during turnover.

The exchange of CO for S²⁻ on the FeMo-cofactor is reversible under turnover conditions.¹⁴³ Thus, CO inhibition of acetylene reduction could be recovered when the CO-bound MoFe protein crystals were dissolved and incubated in an assay mixture, and the structure of MoFe protein crystals obtained from this mixture revealed the loss of the bridging CO ligand and the reinstallation of the belt sulfide, thus restoring the resting state FeMo-cofactor structure. Even though a plausible sulfide binding site (SBS) was proposed, the origin and mechanism of insertion of the belt sulfide ligand are still not clear.

About a decade ago, CO was discovered to be a substrate of V- and Mo-nitrogenases producing both CH₄ and short-chain hydrocarbons as products.^227,228,247 Two or more CO could be reduced and coupled to make multi-C hydrocarbons at low rates when turned over with Fe protein and MgATP. The reactivity of FeMo-co- and FeV-co-reconstituted apo-MoFe protein (apo-NifDK) toward CO reduction has also recently been explored.²⁴⁸ It was found that both FeMo-co- and FeV-co-reconstituted NifDK protein showed similar hydrocarbon product profiles from CO to that of wild-type Mo-nitrogenases, with much lower rates compared to that by V-nitrogenase.^227,228 However, a ~100 times increase of hydrocarbon formation from CO reduction was observed when FeMo-cofactor was housed in a VFe protein scaffold compared to that housed in its native MoFe protein scaffold.²⁴⁹ These results highlighted a combined contribution from the protein environment and cofactor properties toward CO reduction.

It has also been reported that CO can bind to the V-nitrogenase in the absence of Fe protein/MgATP (nonturn-over) as electron-delivery agent, but instead in the presence of the reductants dithionite (E⁰ = −0.66 V), Eu^II-EGTA (E⁰ = −0.88 V), and Eu^II-DTPA (E⁰ = −1.14 V).²⁴³ The comparison of the EPR spectrum of the CO-bound VFe protein to that of the CO-modified MoFe protein led the authors to propose replacement of a belt-sulfide by a bridging CO ligand (Figure 23C).^143,243 They reported that the CO of this CO-bound VFe protein could be further reduced to methane when subjected to turnover condition with Fe protein (VnfH)/MgATP and dithionite, revealing the relevance of this CO-bound state to the catalytic turnover of CO as substrate.²⁴³

It has been long believed that CO does not bind to FeMo-cofactor of Mo-nitrogenase in the resting state (E₀), which is further supported by a recent electrochemical study of CO interaction with FeMo-cofactor in MoFe protein with Eu^III/II-L (L = BAPTA, EGTA, DTPA) as electron mediators.²⁵⁰ However, the binding of CO to FeV-cofactor in the absence of Fe protein/MgATP (resting-state VFe protein as suggested)²⁴³ indicates that it might be risky to simply expand the concept of resting-state E₀ of Mo-nitrogenase to V- and Fe-nitrogenases without careful consideration of the fine-tuning effect of heterometals (Mo, V, and Fe) and secondary environments on the electronic structure of FeM-cofactor. For example, to date, the FeV-cofactor in as-isolated VFe protein always has a CO₃²⁻ ligand that does not exist in any MoFe protein crystal structures observed.^58,59 Clearly, CO interacts with nitrogenases in complex ways, with some aspects still needing resolution.

In 2013, the possible connection between CO reduction and the interstitial carbide in the active site FeMo-cofactor was explored.²⁵¹ It was found that reconstitution of apo-MoFe protein with ¹⁴C-labeled FeMo-cofactor in the resting state and upon turnover with acetylene and N₂ did not show any dilution of the ¹⁴C-label in the active site, indicating that the interstitial carbide cannot be exchanged during turnover. Moreover, the GC-MS characterization of the hydrocarbon products from reaction mixtures with the combined use of ^12/13C-enriched CO substrate and the carbide in the active site revealed that the interstitial carbide could not be used as a substrate and incorporated into the hydrocarbon products.²⁵¹ These results are consistent with a role for carbide in stabilizing the structural integrity of FeMo-cofactor⁷¹ as it proceeds through different conformational states during catalysis.

5.3. CO₂ as Substrate

CO₂ was reported as a substrate for nitrogenase in 1995, being reduced by two electrons to CO.²⁵² The form of CO₂ (CO₃²⁻, HCO₃⁻, and CO₂) that is the substrate remains unclear. Recently, it was reported that in the crystal structures of V-nitrogenase a CO₃²⁻ ligand replaced one of the belt-sulfide ligands of the active site FeV-cofactor (Figure 2).⁵⁸ This proposed CO₃²⁻ ligand is also present in a plausible turnover-relevant structure of the FeV-cofactor,⁵⁹ further complicating the origin of this CO₃²⁻ ligand and the relevance to the CO₂ reducing ability of nitrogenases.

Later studies reported that CO₂ could be reduced by eight electrons/protons to methane (CH₄) by a remodeled Mo-nitrogenase having amino acid substitutions α−70^Val→Ala/α−195^His→Gln.²⁵⁴ In 2016, light-driven in vivo CO₂ reduction to CH₄ by an anoxygenic phototroph, Rhodopseudomonas palustris (R. palustris), which expressed the same remodeled Mo-nitrogenase, was reported.²⁵⁵ This catalytic ability was conferred upon R. palustris by use of a variant of the transcription factor NifA that can activate expression of nitrogenase under all growth conditions.^256–259 More recently, the in vivo CH₄ production from CO₂ reduction has been confirmed to be catalyzed by the wild-type Fe-only nitrogenase from R. palustris and several other nitrogen-fixing bacteria.²⁶⁰ The CH₄ production by the Fe-only nitrogenase in R. palustris was found sufficient to support the growth of an obligate CH_4-utilizing Methylomonas strain. These results suggest that active Fe-only nitrogenase, present in diverse organisms, contributes CH₄ that could shape microbial community interactions.²⁶⁰

Recently, it was discovered that Mo-nitrogenase catalytically reduces carbon dioxide (CO₂) to formate (HCOO⁻) at rates >10 times faster than that of CO₂ reduction to CO and CH₄.²⁵³ This reaction was not inhibited by H₂, indicating that CO₂ reduction does not involve the re/oa equilibrium central to N₂ binding/activation, as expected if the reduction occurs a E_n states with n < 4.²⁵³ In fact, DFT calculations¹³⁶ on the doubly reduced FeMo-cofactor (E₂ state) with a Fe-bound hydride and S-bound proton favor a direct reaction of CO₂ with the hydride (direct hydride transfer reaction pathway, see Figure 24 and 25, lower blue pathway), with facile hydride transfer to CO₂ yielding formate. In contrast, a significant barrier is observed for reaction of Fe-bound CO₂ with the hydride (associative reaction pathway, Figure 25, upper blue pathway), which leads to CO and CH₄. Importantly, computations revealed that protein residues above the reactive face of the FeMo-cofactor hamper the CO₂ access to the hydridic hydrogen. In particular, steric hindrance from the side chain of α−70^Val or α−96^Arg introduces a barrier for the direct hydride transfer favoring the competitive elimination of H₂ over HCOO⁻ formation. Consistent with this finding, MoFe proteins with amino acid substitutions near FeMo-cofactor (α−70^Val→Ala/α−195^His→Gln) are found to significantly alter the distribution of products between formate and CO/CH₄.²⁵³ The formate formation by Fe-nitrogenase reduction of CO₂ was also reported to have a higher efficiency than that of Mo-nitrogenase.²⁶¹

Figure 24.

Possible CO₂ reduction pathways. CO₂ activation at one FeS face of the E₂ state of FeMo-co is shown. The E₂ state is proposed to contain a single Fe-hydride and a proton bound to a sulfide shown bound to one face of FeMo-co. Reduction to formate (blue pathways) can go by either direct hydride transfer or an associative pathway. A pathway to formation of CO and CH₄ is shown in the green. Adapted with permission from ref 253. Copyright 2016 American Chemical Society.

Figure 25.

Computed free energy diagram for CO₂ reduction and H₂ formation occurring at the E₂ state of FeMo-cofactor. Reproduced with permission 253. Copyright 2016 American Chemical Society.

The analysis of the electronic properties of FeMo-cofactor during the hydride transfer revealed a strong charge transfer from the σ(Fe-H) bonding orbital to the π*(C=O) antibonding orbital, which results in the heterolytic Fe-H → Fe⁺ + H⁻ bond cleavage and the transfer of the hydride to CO₂. The exergonic nature of the direct hydride transfer (ΔG⁰ = −26 kcal/mol) suggests that the hydricity of the E₂ state, ΔG_H−, is sufficient to transfer a hydride to CO₂ to generate formate, that is, the hydricity is below that of formate (HCOO⁻ → CO₂ + H⁻, ΔG_H− = +24.1 kcal/mol).²⁶²

The same computational investigation revealed a pathway leading to the formation of CO from E₂(2H)-CO₂, which starts with the exergonic transfer of the protic S-H hydrogen to one of the oxygen atoms of CO₂ (Figure 25, green pathway).²⁵³ Although the small size of the adopted computational models introduces a rather large uncertainty, initial protonation of CO₂ by other ionizable residues, such as α−96^Arg, was found very unlikely. The formation of CO proceeds with the transfer of the Fe-H hydrogen (as a proton rather than a hydride) to that oxygen, which results in the exoergic dissociation of a water molecule.²⁵³

The competitive formation of H₂ from the E₂(2H) state was also investigated computationally (Figure 25, red pathway). It was found that the release of H₂ is a facile and exergonic process (ΔG⁰ = −31.3 kcal/mol). The exergonic nature of the formation of H₂ is consistent with the observation that H₂ cannot reduce FeMo-cofactor. It is of interest to point out that, as with the direct hydride pathway for the CO₂ reduction, the Fe-H behaves as a hydride that is protonated by S-H, with the overall H₂ elimination driven by a σ(Fe-H) to σ*(S-H) charge transfer.²⁵³

In a separate study of CO₂ reduction by nitrogenases, it was found that both Mo- and V-nitrogenases can reduce CO₂ to CO yielding substoichiometric amounts of product compared to the amount of the proteins used.^263,264 Different from its Mo-counterpart, the V-nitrogenase can further reduce CO₂ to CD₄, C₂D_4, and C₂D₆ in D₂O buffer with substoichiometric amounts of products after 3 h incubation. The product profile of CO₂ reduction was further expanded to ≤C₄ hydrocarbons in an ATP-independent study of V-nitrogenase with Eu^II-DTPA as reductant, with all products showing a turnover number (TON) < 1.^264,265

5.4. Other C Substrates

As analogs of carbon dioxide, isocyanic acid/cyanate (HNCO/OCN⁻), thiocyanate (SCN⁻), and carbon disulfide (CS₂) are substrates of wild-type Mo-nitrogenase.^252,266 The slow turnover rates for the reduction of these substrates allowed trapping intermediates by rapid freeze-quench. Dependent on pH and isocyanic acid concentration, the EPR spectra of nitrogenase during turnover in the presence of HNCO displayed signals with g-values corresponding to lo-CO and hi-CO complexes indicating the production of CO.²⁶⁶ The initial EPR study of trapped intermediates from thiocyanate and carbon disulfide reduction by nitrogenase revealed a similar S = 1/2 spin state signal (denoted “c”), indicating a sulfur-containing intermediate bound to FeMo-cofactor.²⁶⁶ Further detailed EPR and ¹³C-ENDOR study²⁶⁷ revealed the presence of three sequentially formed intermediates trapped during CS₂ reduction: “a” with g = 2.035, 1.982, 1.973; “b” with g = 2.111, 2.002, 1.956; and a previously observed “c” with g = 2.211, 1.996, 1.978. All three intermediates were suggested to have CS₂-derived fragments bound to one or two Fe atoms on FeMo-cofactor. The first formed intermediate “a” might contain an activated CS₂ bound to FeMo-cofactor in side-on mode, and the third intermediate “c” was suggested to contain a “C≡S” species terminally bound to the cofactor through C atom.²⁶⁷

Recently, the potential of selenocyanate (SeCN⁻) as nitrogenase substrate and inhibitor was evaluated.¹⁴⁴ Compared to SCN⁻ and cyanide (CN⁻), it was found that SeCN⁻ is a poor substrate in terms of methane production. In addition, SeCN⁻ is a potent, yet reversible inhibitor of acetylene reduction by nitrogenase with an inhibition constant (K_i (SeCN⁻) = 410 ± 30 μM)¹⁴⁴ 30 times lower than that observed for SCN⁻ (K_i (SCN⁻) = 12.7 ± 1.2 mM).²⁶⁶ In contrast to full inhibition of acetylene reduction by SeCN⁻, the proton reduction was inhibited to a lesser extent.¹⁴⁴

Rees et al. resolved a MoFe protein crystal structure at 1.6 Å resolution isolated from an assay mixture that included 25 mM SeCN⁻, revealing a quantitative substitution of the belt sulfide S2B by a selenide ligand.¹⁴⁴ A time course series of crystallographic structures demonstrated the migration of the selenide ligand from the original Se2B position to the other two belt sulfide positions (S5A and S3A) during acetylene reduction, as opposed to little or no migration of Se2B during proton reduction. The Se was completely lost and replaced by S after a sufficient number of catalytic turnovers of acetylene. Turning over of the Se-incorporated MoFe protein in the presence of CO and C₂H₂ resulted in the migration of Se from 2B position to the other two belt positions and high occupancy of bridging CO ligand in the Se2B position (Figure 26).¹⁴⁴ The lability of the S2B position toward ligand exchange during substrate reduction confirms the long-proposed Fe2-Fe6 edge as a primary interaction site for substrates and inhibitors.^{23,33,36,40,59,110,144,156} Application of advanced X-ray absorption spectroscopies to these Se- and CO-substituted FeMo-cofactors revealed a significant asymmetry with regard to the electronic distribution about the cluster, and a redox reorganization mechanism within the cluster has been postulated.²⁶⁸

Figure 26.

Simplified schematic representation of conversion of Se-modified FeMo-cofactor by CO, highlighting the migration of Se²⁻ to S3A and S5A position from the Se2B position.¹⁴⁴

As discussed above, the migration of Se2B into the other two belt positions displayed a reversible and substrate-dependent pattern.¹⁴⁴ This apparent “ratcheting” motion of the selenide in three belt positions on the FeMo-co suggests a dynamic structural change (e.g., the interchange of the belt sulfides through swapping Fe-S partners in the prismatic 6Fe core) of the cluster, while still bound in the protein upon turnover.^144,269 The mechanism of this apparent “ratcheting” motion is not yet clear. On basis of the DFT calculations,¹⁷⁹ the rotation of the prismatic part of the FeMo-co as a whole is less possible. Other pathways involving small molecules, such as S and Se atom carriers, has been suggested and is worth considering.¹⁷⁹ The motion might be substrate specific.²⁶⁹ These studies open the door to further exploration of structure-function relationships of the active site of nitrogenase during nitrogen fixation. Moreover, the mutual substitution reactions between different small molecule ligands, S²⁻, Se²⁻, CO, and possibly acetylene, should be considered in light of the relative bond dissociation energy between different Fe-L bonds, where L = S, Se, or C.

6. CONCLUDING REMARKS

Remarkable progress has been achieved in understanding aspects of the mechanism of nitrogenase reduction of a wide range of substrates, including protons and N₂, nonphysiological N substrates, and C substrates. For N₂ and proton reduction, studies over the last 10 years have provided an understanding of the critical E₄ step where N₂ binds, and H₂ is released. Many lines of evidence have revealed the enzyme functions with the limiting 8e⁻ stoichiometry of eq 1 and Figure 3 because the re and release of H₂ is required to drive one of the most challenging reactions in biology, cleavage of the N≡N triple bond. This exoergic loss of H₂ from E₄(4H), the Janus intermediate drives N₂ binding and formation of the first stage in actual N₂ reduction, the E₄(2N2H) state. The enzyme mechanistically couples exothermic H₂ formation to endothermic cleavage of the N≡N triple bond in the nearly thermoneutral re/oa equilibrium.

One of the tasks remaining is to determine whether there is an alternative pathway, with H₂ re and release at E₃(3H), as also proposed by Lowe and Thorneley. The evidence for this process is not strong, and the question must be re-examined. In the future, it would be useful to develop direct spectroscopic probes of the E_n FeMo-cofactor intermediates with odd-n (even-electron) that could give the types of insight that have been gained using EPR on even-n (odd-electron) species.

The additional questions about the mechanism of nitrogenase reduction of N-substrates include the following. Foremost is unambiguously distinguishing between alternating (A) and distal (D) pathways for N₂ reduction. The fundamental difference between these two pathways is the stage at which the N-N bond is cleaved and the first NH₃ formed: E₅ for D versus E₇ for A. The two trapped reduction intermediates, H and I, do not directly distinguish, as they are respectively E₇ and E₈, stages past N-N cleavage. However, the weight of the evidence provided by studies of the non-native N-substrates points to the A pathway, as described above. Nonetheless, more studies are needed, in particular, trapping and characterization of the odd-electron E₆ state. An initial step in this direction is studies of E₁, and it has been suggested, but not confirmed, that a recent crystal structure of VFe represents E₆ or E₇.⁵⁹

In delivering each [e⁻/H⁺] to FeMo-cofactor, the critical step is the P → M electron transfer that must be accompanied by H⁺ delivery. One might well ask whether this occurs through proton coupled electron transfer (PCET) or coupled electron–proton transfer (CEPT). The ability to observe by EPR both E₇ and E₈ (H and I) provides a chance to study the conversion of H to I by cryoreduction/annealing, as has been done for the E₀ to E₁ conversion.

For nonphysiological N substrates, an important question is understanding how the N species is hydrogenated. Is this through migratory insertion of FeMo-cofactor bound hydride species? Is there any significant structural rearrangement/change that plays important roles during binding/activation/hydrogenation? Answers to such questions will take our understanding of the mechanism to the next level.

Progress in understanding nitrogenase reduction of C-based substrates spans decades, with the most recent advances coming for acetylene, CO, and CO₂. A clearer picture of the mechanism for CO₂ and CO reduction is emerging, with the Fe-hydrides again at the center of the chemistry. Clearly understanding how nitrogenase reduces these C-bound substrates is important in gaining insights into the chemistry that can be achieved at the active site metal clusters. These reactions have also been suggested as a possible avenue for advancing biotechnology for hydrocarbon production. However, the direct use of nitrogenase or its cofactors to contribute to the production of feedstocks or CO₂ sequestration at a useful scale is likely to be an insurmountable challenge given the lack of robustness of the enzymes and their associated cofactors.

Nitrogenase first became available for critical studies in its purified form more than half a century ago.^61,62,270 This was followed by steady progress in understanding the catalytic function of this complex enzyme, with rapid recent progress in determining the enzyme mechanism as described in this Review. Similar progress in addressing the many remaining issues, including those just mentioned, is expected.

Biographies

Lance C. Seefeldt received a B.S. degree in chemistry from the University of Redlands in California and a Ph.D. in biochemistry from the University of California at Riverside. He was a postdoctoral fellow at the Center for Metalloenzyme Studies at the University of Georgia before joining the faculty of the Chemistry and Biochemistry Department at Utah State University, where he is now Professor and Department Head. He was recently awarded the D. Wynne Thorne Career Research Award and was elected a Fellow of the American Association for the Advancement of Science. Over the past 30 years, his research has focused on elucidating the mechanism of nitrogenase.

Zhi-Yong Yang received a B.S. degree in chemistry in 2001 and a Ph.D. degree in organic chemistry in 2006 from Nankai University in Tianjin, China. Following a position at Shanghai ChemPartner as an organic chemist, he joined Prof. Lance Seefeldt’s lab at Utah State University and was awarded with a Ph.D. degree in biochemistry in 2013. After a postdoctoral training in the same group, he was promoted to Researcher in 2017. During the past 19 years, his research has covered biomimetic chemistry of hydrogenases and the mechanism of nitrogenase. His research interest is in understanding the mechanism of small molecule activation catalyzed by biological and synthetic catalysts, including kinetics, thermodynamics, electron transfer, and energy transduction.

Dmitriy A. Lukoyanov received a M.S. degree and a Ph.D. from Kazan State University. He is a Research Associate at Northwestern University. During the last 14 years, his research work has focused on investigation of nitrogenase catalysis with application of various Electron Paramagnetic Resonance (EPR) techniques.

Derek F. Harris received a B.S. in biology from Dixie State University and a M.S. and Ph.D. in biochemistry from Utah State University. He continues to study nitrogenase enzymes as a postdoctoral fellow at Utah State University.

Dennis R. Dean received a B.A. from Wabash College and a Ph.D. from Purdue University. He is currently a University Distinguished Professor at Virginia Tech, where he has also served as the Director of the Fralin Life Sciences Institute, the Virginia Bioinformatics Institute and Vice President for Research. He is a Fellow of the American Association for the Advancement of Science and a Fellow of the American Academy of Microbiology.

Simone Raugei obtained a Ph.D. in theoretical chemistry from the University of Florence (Italy) in 2000 with Prof. Vincenzo Schettino. In 1997–1998, he studied at the Max Planck Institute for Solid State Physics in Stuttgart (Germany) under the supervision of Prof. Michele Parrinello. From 2000 to 2002, he joined the University of Pennsylvania (USA) as a postdoctoral fellow in the group of Prof. Michael L. Klein. From 2002 to 2009, he was an Assistant Professor at the School for Advanced Studies in Trieste (Italy). Since 2010, he has been a senior Scientist at the Pacific Northwest National Laboratory. His current scientific activity focuses on the computational and theoretical modeling of chemical and biochemical processes for energy storage and energy delivery.

Brian M. Hoffman is the Charles E. and Emma H. Morrison Professor of Chemistry and Professor of Molecular Biosciences at Northwestern University. His undergraduate studies were at the University of Chicago, PhD with Harden McConnell at the California Institute of Technology, and Postdoctoral work at the Massachusetts Institute of Technology with Alex Rich. His research in Bioinorganic Chemistry has included measurements of long-range electron transfer between proteins, in addition to determinations of metalloenzyme mechanisms through development and implementation of electron–nuclear double resonance (ENDOR) techniques. He is a member of the National Academy of Sciences and the American Academy of Arts and Sciences.