Identification of a Key Catalytic Intermediate Demonstrates that Nitrogenase is Activated by the Reversible Exchange of N₂ for H₂

Abstract

Freeze-quenching nitrogenase during turnover with N₂ traps an S = ½ intermediate that was shown by ENDOR and EPR spectroscopy to contain N₂ or a reduction product bound to the active-site molybdenum-iron cofactor (FeMo-co). To identify this intermediate (termed here EG), we turned to a quench-cryoannealing relaxation protocol. The trapped state is allowed to relax to the resting E₀ state in frozen medium at a temperature below the melting temperature; relaxation is monitored by periodically cooling the sample to cryogenic temperature for EPR analysis. During −50°C cryoannealing of EG prepared under turnover conditions in which the concentrations of N₂ and H₂ ([H₂], [N₂]) are systematically and independently varied, the rate of decay of EG is accelerated by increasing [H₂] and slowed by increasing [N₂] in the frozen reaction mixture; correspondingly, the accumulation of EG is greater with low [H₂] and/or high [N₂]. The influence of these diatomics identifies EG as the key catalytic intermediate formed by reductive elimination (re) of H₂ with concomitant N₂ binding, a state in which FeMo-co binds the components of diazene (an N-N moiety, perhaps N₂ and two [e⁻/H⁺] or diazene itself). This identification combines with an earlier study to demonstrate that nitrogenase is activated for N₂ binding and reduction through the thermodynamically and kinetically reversible reductive-elimination/oxidative-addition exchange of N₂ and H₂, with an implied limiting stoichiometry of eight electrons/protons for the reduction of N₂ to two NH₃.

Introduction

Biological nitrogen fixation — the reduction of N₂ to two NH₃ molecules — is primarily catalyzed by the Mo-dependent nitrogenase. This enzyme comprises an electron-delivery Fe protein and MoFe protein, a dimer of dimers that contains two copies of the active site FeMo-cofactor (FeMo-co).^1,2 A suggested limiting stoichiometry for nitrogen fixation,³

$${\mathrm{N}}_2+8{\mathrm{e}}^{-}+16\mathrm{A}\mathrm{T}\mathrm{P}+8{\mathrm{H}}^{+}\to 2\mathrm{N}{\mathrm{H}}_3+{\mathrm{H}}_2+16\mathrm{A}\mathrm{D}\mathrm{P}+16{\mathrm{P}}_{\mathrm{i}}$$ (1)

incorporates an obligatory formation of one mole of H₂ per mole of N₂ reduced, and thus a puzzling requirement for two reducing equivalents and four ATP beyond the chemical requirement for N₂ reduction.^1,2 This stoichiometry is embodied in a kinetic framework for nitrogenase function provided by the Lowe-Thorneley (LT) model,^1,2,4 which describes transformations among catalytic intermediates (denoted E_n) where n is the number of electrons and protons (n = 0 – 8) delivered to one half of the MoFe protein (Figure 1A). In this model, N₂ reduction requires activation of the MoFe protein to the pivotal E₄(4H) state (see the kinetic scheme of Figure 1, where the legend explains notation), in which ENDOR has shown FeMo-co to have accumulated four reducing equivalents stored in the form of two [Fe-H-Fe] bridging hydrides,^5–7 presumably with two protons bound to sulfides of FeMo-co (Figure 1B, left).^5,7–9 We recently proposed^{8, 10} and subsequently provided experimental evidence¹¹ that nitrogenase is activated for N₂ binding and reduction through reductive elimination (re)^12–15 of the two bridging hydrides of E₄(4H) to form H₂ (Figure 1B), thereby experimentally supporting the stoichiometry of eq 1.

Figure 1.

(A) Simplified LT kinetic scheme highlighting correlated electron/proton delivery in eight steps, with inclusion of some of the possible pathways for decay by H₂ release; although N₂ binds at either the E₃ or E₄ levels, only the E₄ state is reactive, so the pathway through the E₄ state is emphasized. For states under discussion in this report, parentheses added to the E_n notation to denote the stoichiometry of H/N bound to FeMo-co. When the molecular species corresponding to this stoichiometry is specified, it is noted; for example, E₄(2N2H) might have N₂ or N₂H₂ bound, and would be notated as E₄(N₂,2H) and E₄(N₂H₂). (B) The reductive-elimination (re) mechanism for H₂ release upon N₂ (blue) binding to E₄(4H) and its reverse, oxidative-addition (oa) of H₂ with loss of N₂ (eq 4) visualized as occurring on the FE2,3,6,7 face of FeMo-co. The binding modes of the hydrides of E₄(4H) and the components of diazene in E₄(2N2H) are arbitrary.

As part of the development of the re mechanism we used advanced paramagnetic techniques to characterize two nitrogenous E_n intermediates that are associated with states in the LT scheme subsequent to N₂ binding, n ≥ 4 (Figure 1), and that are common to turnover of remodeled nitrogenase with N₂H₂, Me-N₂H, N₂H₄, NO₂⁻, and H₂NOH. One is a non-Kramers (S ≥ 2) state, denoted H, that is assigned as E₇, with bound [-NH₂]; the second is a Kramers (S = ½) state, denoted I, assigned as E₈, with bound NH₃.^{8, 10,16} During turnover of wild-type nitrogenase with N₂ an additional intermediate state with S = ½ was trapped, but not identified.^17–19 With a half-integer spin, like the E₀ resting state, this state must differs from E₀ by the accumulation of the delivery of an n = even number of [e⁻/H⁺] to FeMo-co, This state, herein denoted as EG, must further correspond to an E_n state formed subsequent to binding N₂, n = 4, 6, or 8, because it gives ¹⁵N ENDOR signals when trapped using ¹⁵N₂ substrate. The previous assignment of I as E₈, a product (NH₃)-bound state,^{8, 10} implies an assignment of EG to E₄ or E₆. The notation, EG, in fact was adopted because if the states of Figure 1A were to be labeled sequentially by letters beginning with E₀ = A, then E₄ would be E and E₆ would be G.

Because of the relatively low accumulation of EG in freeze-quenched samples, to date neither ENDOR measurements nor the use HYSCORE, as successfully applied to intermediate I,²⁰ have yielded an assignment of EG. To identify EG, we therefore here turn to a quench-cryoannealing relaxation protocol developed to determine the reduction level, n, of an EPR-active freeze-trapped E_n intermediate state.⁶ The trapped state is allowed to relax to the resting E₀ state in frozen medium, at temperatures T ≤ −20°C, well below the melting temperature of the buffered samples, T ~ 0°C. Keeping the sample frozen prevents any additional accumulation of reducing equivalents, because binding of reduced Fe protein to and release of oxidized protein from the MoFe protein both are abolished in a frozen solid. As recently confirmed,²¹ the frozen intermediate can relax towards the resting state only through steps that release a stable species from FeMo-co. The E_n states formed prior to N₂ binding lose two equivalents per relaxation step by releasing H₂: E₂(2H) relaxes to E₀ with loss of H₂; E₃(3H) can relax to E₁(1H) by loss of H₂, but as there is no center available to accept a single electron, E₁(1H) cannot further relax to E₀;²¹ E₄ is a special case with multiple substates (Figure 1B), which will be discussed. States formed subsequent to N₂ binding/reaction, n ≥ 4, would relax through release of a stable form of reduced substrate. As the defining example of this approach, the freeze-trapped S = ½ intermediate shown by ENDOR spectroscopy to exhibit two [Fe-H-Fe] bridging hydrides^5,7 was identified as E₄(4H) by its relaxation to the resting state E₀ through the release of a total of four reducing equivalents in a two-step process, each step releasing H₂ (two equivalents), with formation of E₂(2H) in the first step (Figure 1A).⁶

Application of this quench-cryoannealing relaxation protocol to EG identifies its E_n turnover state as one in which the N-N bond is intact. Most importantly, the results combine with our earlier study¹¹ to directly demonstrate that nitrogenase activation occurs as a thermodynamically and kinetically reductive-elimination/oxidative-addition exchange of N₂ and H₂ at the E₄ reduction level, as represented by Figure 1B.

Materials and Methods

General procedures

All chemicals were obtained from Sigma-Aldrich Chemicals, and were used without further purification. Hydrogen, argon, and dinitrogen gases were purchased from Air Liquide America Specialty Gases LLC (Plumsteadville, PA). The argon and dinitrogen gases were passed through an activated copper-catalyst to get rid of contamination of oxygen before use. Azotobacter vinelandii strains DJ995 (wild type MoFe protein) and DJ884 (wild type Fe protein) were grown and nitrogenase proteins were expressed and purified as previously described.²² Both proteins were greater than 95% pure as confirmed by SDS-PAGE analysis using Coomassie blue staining, and were fully active, as discussed in SI. Proteins and buffers were handled anaerobically in septum-sealed serum vials under an inert atmosphere (argon or dinitrogen) or on a Schlenk vacuum line. All transfers of gases and liquids were done with Hamilton gastight syringes.

Preparation of Cryoannealing samples

Samples were prepared by adding Fe protein and MoFe protein to a solution containing a MgATP regeneration system, such that the final concentrations were 13 mM ATP, 15mM MgCl₂, 20 mM phosphocreatine, 2.0 mg/mL bovine serum albumin, and 0.3 mg/mL phosphocreatine kinase (200 mM MOPS buffer at pH 7.3 with 50mM dithionite). MoFe protein was added first to a final concentration of ~50 µM; reaction was then initiated by the addition of Fe protein to a final concentration of ~75 µM. After 20–25 sec incubation at room temperature, ~300 µL of the reaction mixture (total volume ~ 400 µL) was transferred into 4-mm calibrated quartz EPR tubes and rapidly frozen in a hexane slurry before being stored in liquid nitrogen for EPR analysis. When effects of ammonium ion were studied, reaction mixtures contained a final ammonium chloride concentration of 0.2 mM.

To prepare unstirred samples with a selected N₂ partial pressure under normal conditions, N₂ (or Ar) gas was added to an argon (or N₂)-filled 9.4-mL assay vial with a 500 µL Eppendorf tube inside. The final pressure in all vials was 1 atm. When the stirring effect was studied, the assay vial contained a small magnetic stirring bar, the final volume of liquid phase was doubled to 0.8 mL, and the reaction mixture was stirred to equilibrate the gas and liquid phases. To study the effect of gas flushing in addition to stirring, the gas mixture with the chosen N₂ partial pressure in a 60-mL syringe was flushed through the headspace of the reaction vial from a syringe inserted through the water-sealed septum and ventilated through a second water-sealed small syringe attached through the septum; a total volume of 30 mL of gas mixture was flushed through the reaction vial prior to freezing. The NH₃, H₂ and N₂ concentrations in frozen samples prepared with the two protocols, unstirred and stirred, were estimated as described in SI (Figures S1–S3).

Cryoannealing EPR methods

The cryoannealing protocol has been described.^{6, 21} It involves multiple steps in which a freeze-quenched sample at liquid nitrogen temperature is rapidly warmed to the annealing temperature by immersion in a methanol bath held at that temperature, annealed at that temperature for a fixed time, then quench-cooled back to 77 K by immersion into liquid nitrogen. CW X-band EPR is then collected on an ESP 300 Bruker spectrometer equipped with Oxford ESR 900 cryostat.

A freeze-trapped intermediate, E_n, typically decays with a distribution of rate constants, which can be described with a stretched exponential function, eq 2,²³

$$E_n(t)\propto \text{exp}(-{(t/\tau )}^m)$$ (2)

where τ is the average decay time and 0 ≤ m ≤ 1 reflects the breadth of the distribution, with m decreasing as the breadth of the distribution increases.²⁴ In formulating the differential equations of a multi-step process, the distribution of rate constants that characterizes a stretched exponential decay is manifest as a time-dependent rate constant, k(t).^{6, 24}

Results and Discussion

Figure 2 (t = 0) presents the EPR spectrum of nitrogenase quench-frozen during steady-state turnover (ca. 25 sec after mixing) under an N₂ partial pressure (P(N₂)) of 0.05 atm. At low fields it shows the g₁/g₂ features of S = 3/2 signals, and is decomposed as previously described²¹ (also see legend to Figure 3) into a contribution from the E₀ resting-state FeMo-co (signal denoted 1a), and a larger contribution from the E₂(2H) intermediate (denoted 1b).^{6, 21} The g-2 region is dominated by the spectrum for the S = ½ EG intermediate, g = [2.08, 1.99, 1.97].

Figure 2.

Selected EPR spectra collected during the time course for −50°C cryoannealing of a sample freeze-trapped under P(N₂) = 0.05 atm, and showing the decompositions of the S = 3/2 spectra in the low-field g₁g₂ region into contributions from 1a (g = [4.32, 3.66, 2.01], red) and 1b (g = [4.21, 3.76, ~1.97], blue). At early time the g-2 region is dominated by EG, as indicated. The indicated signal from the reduced 4Fe/4S cluster of Fe protein is greatly distorted at this temperature, but it is clear that its intensity does not change with annealing. EPR conditions: temperature, 3.8 K; microwave frequency, 9.36 GHz; microwave power, 0.5 mW; modulation amplitude, 13 G; time constant, 160 ms; field sweep speed, 38 G/s.

Figure 3.

(A) Decay of intermediate EG freeze-trapped during turnover of nitrogenase under P(N₂) = 0.05 (open circles; see Figure 2) and 1 atm (solid circles) observed during cryoannealing at −50 °C. Plotted are intensities of g₁ feature of the S = ½ EPR signal, shown after normalization to the maximum (zero-time) signal and fitted with a stretched exponential decay function, eq 2. EPR conditions: as for Figure 2, except for field sweep speed of 20 G/s. (B) Progress curves for the three EPR-active species, EG, 1a, and 1b, observed during cryoannealing of EG freeze-trapped under P(N₂) = 0.05 atm. Fits to kinetic scheme (eq 3) as previously described; parameters of k₁ for EG stretched-exponential decay presented in 3A; rate of the second step of eq 3, k₂ = 0.0023 min⁻¹.^{6, 21} Intensities for the resting state (black) and E₂ state (blue) obtained with the previously described²¹ procedure of deconvolution and quantitation of corresponding S = 3/2 EPR signals 1a and 1b (see Figure 2). Intensities for the EG intermediate (red) taken from Figure 3A and converted to concentration units by scaling to the two-step kinetic scheme for decay, eq 3. EPR conditions: as for Figure 3A, except for field sweep speed of 38 G/s for 1a and 1b signals detection.

Figure 2 further shows representative spectra collected during −50 °C cryoannealing of the freeze-trapped sample; the EG decay at higher annealing temperatures, (e.g., −20 °C) was too rapid to monitor conveniently. As shown, the annealing leads to progressive loss of the EG signal and increase in the E₀ signal, plus a rise then fall of the E₂ signal.

Figure 3A presents the progress curves for −50 °C cryoannealing of samples trapped during steady-state turnover both under (P(N₂)) of 0.05 and 1 atm. As is commonly the case in cryoannealing,^{6, 23} the decay of EG formed under a partial pressure of P(N₂) = 1 atm is best described with a stretched exponential,²⁴ eq 2, indicating that there is a distribution of rate constants associated with an ensemble of slightly differing conformations of the intermediate trapped in the frozen solution. The decay at this temperature is slow, with an average decay time of τ = 460 min and a ‘stretch’ constant, m = 0.66. Strikingly, when EG is trapped during turnover under an atmosphere of only P(N₂) = 0.05 atm, the decay dramatically speeds up: the average decay time decreases more than twelve-fold, to τ = 40 min; moreover, the decay becomes nearly exponential, m = 0.93.

As shown in Figure 2 and illustrated in 3B, decay of EG during cryoannealing leads to the appearance of the S = 3/2 signal, denoted 1b, associated with the E₂(2H) intermediate.^{6, 21} This species in turn decays with the gradual recovery of resting state E₀. As ENDOR measurements show that EG is a state that has bound N₂,^{8, 10,16} and this state decays through E₂(2H) to E₀ during cryoannealing, then according to Figure 1A EG can only be E₄(2N2H). According to this kinetic model, E₄(2N2H) relaxes via E₄(4H) and E₂(2H) to resting state E₀ with loss of N₂ and two H₂. This is described by Scheme 1, which extracts from Figure 1A those states associated with relaxation of E₄(2N2H) to resting E₀. Multiple repetitions of the experiment, have shown that in general low amounts of an intermediate whose EPR signal can be assigned to E₄(4H) are freeze-trapped, and that little-to-no additional E₄(4H) accumulates during annealing. As a result of the latter, in particular, the curves for all three observed species can be described jointly⁶ by the phenomenological two-step sequential kinetic scheme derived from Scheme 1 under the condition of minimal accumulation of E₄(4H), eq 3;

$$E_4(2N2H)\stackrel{k1}{\to }{E}_2(2H)\stackrel{k2}{\to }{E}_0$$ (3)

This implies a steady-state approximation for the concentration of E₄(2N2H), presented below; the relationship between the observed EG/E₄(2N2H) decay constants, k₁, k₂, of eq 3, and the microscopic rate constants of Scheme 1 are derived there.

Scheme 1.

Kinetic scheme for the decay of freeze-trapped E₄(2N2H) derived from Figure 1A; k_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 irreversible¹⁰ decay of E₄(4H) and E₂(2H), respectively.

The progress curves for the three species, EG/E₄(2N2H), 1b/E₂(2H), and 1a/E₀, are indeed well-fit by the kinetic relaxation scheme of eq 3, as illustrated for the sample trapped under P(N₂) = 0.05 atm (Figure 3B). In agreement with this scheme, the appearance of 1b/E₂(2H) follows the stretched-exponential (eq 2) associated with the decay of EG, with changes in the partial pressure P(N₂) causing the sharp changes in the average decay time and stretch factor that characterize k₁, as presented above. Correspondingly, 1a/E₀ appears with the rate constant associated with relaxation of 1b/E₂(2H). In contrast, one expects the relaxation of E₂(2H) to E₀ to be independent of turnover conditions,^{6, 21} and this is so. At both values of P(N₂), the process can be described by an exponential with the same rate constant, implying that k₂ = k_d’ of Scheme 1.

The assignment of EG as E₄(2N2H) has been tested, and confirmed, by experiments inspired by the question, why does increasing the concentration of the substrate [N₂] in the frozen solution slow the decay of E₄(2N2H) (Figure 3A), which led to the further questions, what are the possible influence(s) on EG decay of the turnover products, NH₃ and H₂? We first tested whether the NH₃ product (NH₄⁺ at pH 7.3) is the causative agent, perhaps because it binds to MoFe in the vicinity of the FeMo-co of E₄(2N2H) and stabilizes the intermediate, for example by H-bonding to a partially reduced N₂. As NH₃ production increases with increasing P(N₂) (Figure S1), such an effect would be enhanced at high P(N₂), thereby decreasing the rate of decay, as observed.

To test this hypothesis, we first quantified the ammonia concentration in the samples prepared under the freeze-quench turnover conditions used for making the EPR samples in Figure 3. The total concentration of ammonia (NH₄⁺ at pH 7.3) generated under 1 atm of N₂ is about 0.15 mM; that produced under 0.05 atm of N₂ is about four to five fold less (Figure S1). We then freeze trapped B during turnover under 0.05 atm of N₂ in the presence of 0.2 mM of externally added NH₄⁺, even more than the amount produced at 1 atm of N₂. There is no previous report of which we are aware that suggested the presence of NH₄⁺ alters catalysis, and we find that added NH₄⁺ has no influence on either the intensity of the EPR signal from trapped EG or the cryoannealing kinetics of EG (not shown). Hence we discard this possibility.

An influence of the diatomics N₂ and H₂ on E₄(2N2H) decay is implicit in our mechanistic proposal that E₄(2N2H) is formed by N₂ binding and H₂ release through re, Figure 1B. According to this mechanism, decay of EG would occur by the reverse of the re equilibrium, (Scheme 1) and would involve reaction with the H₂ formed during turnover with release of N₂ to generate E₄(4H). Hence, decay would be enhanced by increasing [H₂]; the E₄(4H) thus formed by reaction with H₂ would in turn decay to E₀ through the successive release of two H₂. However, the E₄(4H) can re-react with N₂ in the frozen solution to regenerate E₄(2N2H), so increasing [N₂] would suppress the decay. Thus, according to this reversible-re scenario, varying the concentrations [H₂] and [N₂] in frozen samples would competitively modulate both the accumulation and annealing of EG.

According to this hypothesis, decay of E₄(2N2H) is slower in samples prepared under high P(N₂), Figure 3, because the E₄(4H) state formed during E₄(2N2H) decay undergoes enhanced regeneration back to E₄(2N2H) through reaction with the high solution concentration of N₂ according to Scheme 1. Indeed, the measurements of [N₂] in solution, Figure S3, show that [N₂] in the solution prepared at P(N₂) = 1 atm is more than an order of magnitude higher than that at 0.05 atm. This in turn correlates with the more than ten-fold slower decay of EG in the sample prepared under high P(N₂), consistent with the expectation E₄(2N2H) decay is being suppressed by enhanced re-reaction of E₄(4H) with dissolved N₂ to regenerate E₄(2N2H). One might imagine that the difference in decay rates of Figure 3 reflects the reaction of E₄(2N2H) with differing concentrations of H₂ formed during turnover under the different values of P(N₂). However, experiments described in SI argue against this alternative. Although more H₂ is produced during turnover under low P(N₂),^1,2,3 saturating concentrations of H₂ are produced by the enzymatic reaction under all partial pressures of N₂ used (Figure S2).

To test the reversible-re hypothesis in experiments where both [H₂] and [N₂] are actively varied, we complemented our standard procedures for sample preparation with the simple expedient of flushing the headspace with an N₂/Ar mixture of selected P(N₂) while rapidly stirring, thereby sweeping out the enzymatically produced H₂ into the headspace over the reaction mixture and generating samples with low [H₂], while ensuring equilibration of the reaction mixture with the selected P(N₂) in the headspace gas phase. Stirred and unstirred sample pairs prepared with the same P(N₂) have essentially the same [N₂], but the unstirred samples have high (saturating) [H₂], while the stirred samples have low [H₂].

Figure 4 presents the results of −50°C cryoannealing of two stirred/unstirred pairs of samples of intermediate EG freeze-trapped during steady-state turnover. One pair was under P(N₂) = 0.1 atm and 0.9 atm of Ar; the second pair was freeze-trapped under P(N₂) = 1 atm. Comparison of decay curves for sample pairs with high vs low P(N₂), both the stirred and the unstirred pairs, again shows that high [N₂] suppresses E₄(2N2H) decay, regardless of whether the solution contains high [H₂] (unstirred) or low (stirred).²⁵ Conversely, the effects of stirring on samples prepared under equal P(N₂) show that the higher [H₂] in the unstirred partner of a pair with identical P(N₂) speeds the decay, regardless of whether [N₂] is high (P(N₂) = 1 atm) or low (0.1 atm).

Figure 4.

Progress curves for −50 °C cryoannealing of EG intermediates formed during turnover under P(N₂) = 0.1 (open circles) and 1 (closed circles) atm, with (red) or without (black) stirring reaction mixture. Blue arrows connect pairs of samples prepared with different P(N₂), with arrow pointing towards higher solution [N₂], namely P(N₂) = 1 atm; within a pair, [H₂] is comparable, either low because in both samples enzymatically produced H₂ has been flushed out by stirring, or high in unstirred samples; green arrows connect pairs of samples prepared under the same P(N₂), and thus with comparable [N₂] (either low because P(N₂) = 0.1 atm or high, P(N₂) = 1 atm) but different [H₂], with arrow pointing from sample in which [H₂] is low because stirring has flushed out the H₂ into the headspace, towards unstirred sample with higher [H₂]. Intensities measured as described in Figure 3A. Decays are fit as stretched exponentials (eq 2) with the parameters in the following table:

Turnover conditions	P(N2) = 0.1 atm		P(N2) = 1 atm
	Unstirred	Stirred	Unstirred	Stirred
τ (min)	19	142	224	667
m	0.94	0.71	0.68	0.55

Examination of Figure 1 shows that these effects clearly identify EG as E₄(2N2H). Considering the other E_n, n ≥ 4, even, states, as noted above, E₈ has already been identified with intermediate I, and in any case would not respond to changes in [N₂] or [H₂]. Whether E₆, which we have assigned as binding N₂H₄,^{8, 10} decayed by loss of N₂H₄ or through the release of N₂H₂ and H₂, it would not respond to changes in [N₂], and increasing [H₂] would not speed its decay. Moreover, even in an alternative reaction pathway that has been discussed,^{8, 10} where E₆ contains a moiety generated by cleavage of the N-N bond, here too its decay could not be influenced by changes in [N₂] or [H₂]. Instead, Figure 1 shows that the effects of the two diatomics clearly require not only that EG is E₄(2N2H), but also that its decay, Figs 3, 4, is to be understood as the reverse of the re activation equilibrium,

$${\mathrm{E}}_4(4\mathrm{H})+{\mathrm{N}}_2\rightleftharpoons {\mathrm{E}}_4(2\mathrm{N}2\mathrm{H})+{\mathrm{H}}_2$$ (4)

which leads to the decay of E₄(2N2H) through Scheme 1: the oxidative-addition reaction of E₄(2N2H) with H₂ and loss of N₂ to form E₄(4H) (k_b). The competing regeneration of E₄(2N2H) from E₄(4H) through reaction with N₂ (k_r) and reductive elimination of H₂, in turn competes with the irreversible relaxation of E₄(4H) through loss of H₂ (k_d). This competition is captured by a steady-state treatment of the concentration of E₄(4H) within the kinetic relaxation Scheme 1, which generates as the functional form for k₁(t), the observed rate constant of E₄(2N2H) relaxation according to eq 3,²⁶

$$k_1(t)=\left(\frac{k_d{k}_b[H_2]}{k_r[N_2]+k_d}\right)$$ (5a)

In the limit of eq 5a where k_d << k_r[N₂], the two states E₄(2N2H) and E₄(4H) are in an equilibrium controlled by the ratio, [H₂]/[N₂]; the equilibrium population decays to E₂(2H) with loss of H₂, with rate constant k₁(t) represented by the limiting form of eq 5b,

$$k_1(t)\approx \left(\frac{k_d{k}_b[H_2]}{k_r[N_2]}\right)\propto \frac{[H_2]}{N_2}$$ (5b)

in agreement with the behavior exhibited in the present experiments, Figure 4. The rate constant k₁(t) is expected to be time-dependent because the distribution of rate constants that characterizes a stretched-exponential decay in the frozen solid²³ is manifest as a time-dependent rate constant;²⁴ time-dependence of the concentrations of the diatomics during annealing of the frozen solutions would also contribute. If conditions can be established such that E₄(4H) is trapped in quantities that allow k_d for wild-type nitrogenase to be measured directly, then the equilibrium constant of re (Figure 1B, eq 4), (k_r/k_b) in Scheme 1, can be estimated from eq 5b.

We also studied the effects of varying [H₂] and [N₂] on the amount of EG/E₄(2N2H) trapped through the preparation of paired samples, with and without stirring/flushing, under a range of N₂ partial pressures, Figure 5. At the low partial pressure of P(N₂) = 0.2 atm, stirring/flushing to remove H₂ increases the concentration of trapped EG by ~4–5 fold. However, as the partial pressure of N₂ substrate rises, the significance of stirring/flushing drops, and by 1 atm N₂ the increase is less than two-fold. The improved yield of EG by stirring/flushing effect can be easily explained by the effect of changes of [H₂] and [N₂] on the competition of H₂ with N₂ for binding to the different E₄ substates, as shown in Figure 1B and eq 4. Under all partial pressures employed, a saturating level of H₂ is generated in the unstirred sample (Figure S2), so the removal of H₂ by stirring/flushing would have (roughly) the same influence for all samples. However, the majority of dissolved N₂ is used up before freezing when P(N₂) is low, but there is considerable residual N₂ at the time of freezing for high P(N₂) (Figure S3). Thus, the equilibration of the gaseous and solution N₂ has a lesser effect at high P(N₂).

Figure 5.

Variation in the EPR amplitude of trapped intermediate EG g₁ feature at various N₂ partial pressures when reaction mixture is neither stirred nor flushed (red); stirred (green); stirred with the headspace of reaction vial flushed with appropriate N₂/Ar mixture during turnover. EPR conditions: as described in Figure 3A.

Conclusions

EPR/ENDOR spectroscopy has shown that the S = ½ intermediate EG freeze-trapped during turnover of nitrogenase under N₂ contains N₂ or a reduction product bound to FeMo-co.^17,18,19 This implies that the intermediate must be an E_n state with n = 4, 6, or 8. Cryoannealing experiments on EG prepared under turnover conditions in which the concentrations, [H₂], [N₂] are systematically varied demonstrate that reaction of EG with the H₂ produced during turnover enhances the decay of EG and limits its accumulation, whereas reaction of N₂ with E₄(4H) formed during the decay process slows the observed decay of EG. In the kinetic scheme of Figure 1A there is only one such state that could be affected by H₂, namely E₄(2N2H), the key catalytic intermediate formed upon reductive elimination of H₂ and binding of N₂ (Figure 1B), whose decay involves the [H₂], [N₂] diatomics through the process visualized in Scheme 1. EG/E₄(2N2H) is a state in which FeMo-co binds the components of diazene, which may be present as N₂ and two [e⁻/H⁺] or as diazene itself, an issue to be addressed by future ENDOR measurements. Overall, however, the freeze-quench strategy nonetheless has trapped three of the five intermediate states in which N₂ is involved: one of the substates of E₄, E₇, E₈,^5,7–9 leaving only E₅, E₆ unexamined.

Of primary importance, the present finding that decay of EG is accelerated by increasing [H₂] and slowed by increasing [N₂] in the frozen reaction mixture, and the resulting identification of EG as E₄(2N2H) directly demonstrates that activation of nitrogenase for N₂ binding and reduction involves the thermodynamically and kinetically reversible reductive-elimination/oxidative-addition (re/oa) activation equilibrium of Figure 1B and eq 4, as previously inferred,¹¹ with its implied limiting stoichiometry of eight electrons/protons for the reduction of N₂ to two NH₃ (eq 1).