A methyldiazene (HNNCH₃)-derived species bound to the nitrogenase active-site FeMo cofactor: Implications for mechanism

Abstract

Methyldiazene (HNNCH₃) isotopomers labeled with ¹⁵N at the terminal or internal nitrogens or with ¹³C or ²H were used as substrates for the nitrogenase α-195^Gln-substituted MoFe protein. Freeze quenching under turnover traps an S = ½ state that has been characterized by EPR and ¹H-, ¹⁵N-, and ¹³C-electron nuclear double resonance spectroscopies. These studies disclosed the following: (i) a methyldiazene-derived species is bound to the active-site FeMo cofactor; (ii) this species binds through an [-NH_x] fragment whose N derives from the methyldiazene terminal N; and (iii) the internal N from methyldiazene probably does not bind to FeMo cofactor. These results constrain possible mechanisms for reduction of methyldiazene. In the Chatt–Schrock mechanism for N₂ reduction, H atoms sequentially add to the distal N before N-N bond cleavage (d-mechanism). In a d-mechanism for methyldiazene reduction, a bound [-NH_x] fragment only occurs after reduction by three electrons, which leads to N-N bond cleavage and the release of the first NH₃. Thus, the appearance of bound [-NH_x] is compatible with the d-mechanism only if it represents a late stage in the reduction process. In contrast are mechanisms where H atoms add alternately to distal and proximal nitrogens before N-N cleavage (a-mechanism) and release of the first NH₃ after reduction by five electrons. An [-NH_x] fragment would be bound at every stage of methyldiazene reduction in an a-mechanism. Although current information does not rule out the d-mechanism, the a-mechanism is more attractive because proton delivery to substrate has been specifically compromised in α-195^Gln-substituted MoFe protein.

The reduction of dinitrogen (N₂) to two ammonia (NH₃) molecules (N₂ fixation) represents the exclusive pathway for input of N₂ into the global biogeochemical N cycle and therefore is a reaction essential to all life (1). The majority of all N₂ fixation occurs by the action of microbes that contain the enzyme nitrogenase. The Mo-based nitrogenase catalyzes the reduction of N₂ to two ammonia molecules (2) according to the optimal reaction shown in Eq. 1, requiring multiple electrons, protons, the hydrolysis of MgATP, and the evolution of H₂.

N₂ binding and reduction by nitrogenase occurs at a complex metal cofactor contained in the MoFe protein component protein called FeMo cofactor (3, 4). The structure of this active-site cofactor has been disclosed by the high-resolution x-ray crystal structures of the MoFe protein by the Rees group (5–8) and others (9, 10) (Fig. 1). Recent studies using combined genetic, biochemical, and spectroscopic approaches have localized a binding site for N₂ and nonphysiological alkyne substrates to a specific Fe-S face of FeMo cofactor (11) defined by Fe atoms 2, 3, 6, and 7 (numbering based on Protein Data Bank ID code 1M1N). Further, a strategy has been devised to trap intermediate states during the reduction of the alkyne substrates propargyl alcohol (12) and acetylene (13, 14) in high concentration and to show by electron nuclear double resonance (ENDOR) spectroscopy that they contain reduced substrate bound to FeMo cofactor (13–15). The ENDOR spectroscopic studies reveal that the product alkene most likely is bound side-on to a single Fe atom of FeMo cofactor. Such side-on binding of alkenes (and alkynes) to metals is consistent with binding modes observed in metal complexes outside of proteins (16). An analogous study has characterized an intermediate trapped during proton reduction (17).

Fig. 1.

A view of the FeMo cofactor highlighting Fe atoms 2, 3, 6, and 7 (numbering from Protein Data Bank ID code 1M1N), along with MoFe protein amino acids ligands (α-275^Cys and α-442^His) and the side chains of α-195^His and α-70^Val. The figure was generated by using the Protein Data Bank ID code 1M1N. Fe is green, S is yellow, Mo is purple, C is gray, O is red, and N is blue.

The mechanism for N₂ reduction to ammonia by nitrogenase and the nature of the N₂ reduction intermediates bound to FeMo cofactor during this reaction remain obscure (11). In contrast, a mechanism for N₂ reduction at mononuclear Mo metal complexes is well advanced, with contributions from the early work of Chatt et al. (18) and Pickett (19) and more recently from Yandulov and Schrock (20) and Schrock (21, 22). Determination of the mechanism of N₂ reduction by such Mo-based complexes included the structure determination of a number of intermediates along the reaction pathway. By analogy to the mechanism of these inorganic metal complexes, there is evidence that N₂ reduction catalyzed by nitrogenase proceeds through intermediate states generated by the sequential addition of protons and electrons to N₂ bound to FeMo cofactor (23), forming nitrogenous species at the reduction levels of diazene and hydrazine (Eq. 2, where M represents FeMo cofactor).

In support of such a mechanism is the observation that hydrazine is both a nitrogenase substrate (24, 25), being reduced to ammonia, and a minor product that can be isolated when nitrogenase is acid-quenched during N₂ reduction (23).

We recently reported a strategy for trapping states with bound species when N₂, a diazene, or hydrazine, are used as substrate (26). Key to trapping certain of these species in high occupancy was the use of MoFe proteins that contain amino acid substitutions, separately, or in combination, that compromise proton delivery to nitrogenous substrates (α-195^Gln) or expand accessibility to the substrate reduction site (α-70^Ala). EPR and ENDOR analysis of these trapped states revealed that each of them represents either the initial substrate or a reduced species bound to FeMo cofactor and indicated that at least two, and possibly three, distinct states could be trapped (26). We have suggested that these species could be analogous to different stages along the N₂ reduction pathway. Here, we describe a multinuclear ENDOR analysis of a species that was trapped during nitrogenase catalysis when ¹⁵N- and ¹³C- or ²H-labeled methyldiazene isotopomers were used as substrate. These studies expand our initial insight into the nature of the methyldiazene-bound state and present powerful constraints on possible mechanisms of its reduction by the α-195^Gln MoFe protein.

Results

Trapping Nitrogenous Species Bound to FeMo Cofactor.

We have sought to gain insight into the nitrogenase mechanism by using freeze-quench experiments to trap nitrogenous species bound at the active site when N₂ and other nitrogenous compounds that could be considered as intermediates or analogs in the reduction of N₂ are used as substrates (25, 26). An important aspect in the development of this strategy was the prediction that restricting proton delivery to substrates could result in the accumulation of turnover states with enzyme-bound semireduced species. This prediction was supported by freeze-quench experiments where an S = ½ state was trapped when hydrazine was used as a substrate for the α-195^Gln-substituted MoFe protein (25). The α-195^Gln-substituted MoFe protein appears to be specifically compromised in its ability to effectively deliver protons to nitrogenous substrates (27, 28). The accumulation of this state could be further increased (25) by placing another substitution (α-70^Ala) within the MoFe protein that lowers constraints on the accessibility of substrates to the active site (Fig. 1). Our next goal was to trap a state earlier along the N₂ reduction pathway by using a less-reduced nitrogenous compound as substrate. In the present work we elected to use methyldiazene as substrate because it is asymmetrical and such asymmetry could be exploited to enable discrimination of binding modes by using specifically labeled ¹⁵N or ¹³C isotopomers in freeze-quench ENDOR experiments.

The synthesis and properties of methyldiazene were described some years ago (29). Of importance for the current studies, methyldiazene can be synthesized (Scheme 1) from readily available starting materials with appropriate ^14/15N or ^12/13C labels to produce the requisite isotopomers. The short half-life of methyldiazene in solution necessitated special handling procedures that typically involved synthesis of methyldiazene immediately before it was needed, stabilization by freezing, and rapid transfer of the gas directly to the nitrogenase assay vessel.

Scheme 1.

Methyldiazene synthetic scheme. Hydroxylamine-O-sulfonic acid (3) was synthesized as a white precipitate after reaction of chlorosulfonic acid (1) and hydroxylamine (2, where the N atom was either ¹⁵N or ¹⁴N). N-methylhydroxylamine (5) was synthesized by the reduction of nitromethane (4, where the N atom was either ¹⁴N or ¹⁵N and the C atom was either ¹²C or ¹³C). Methyldiazene (6) was prepared by the base-catalyzed reaction of hydroxylamine-O-sulfonic acid (3) with N-methylhydrolyamine (5), and the gas was isolated by freezing. See Materials and Methods for details.

The instability of methyldiazene and accumulation of its nonenzymatic degradation products precludes a direct demonstration that methyldiazene is an effective nitrogenase substrate. Interaction of methyldiazene with the nitrogenase active site was therefore established by its capacity to inhibit nitrogenase-catalyzed H₂ evolution (30). In the absence of other substrates, nitrogenase reduces protons yielding H₂. The presence of substrates such as N₂ or acetylene diverts electron flow away from proton reduction, resulting in lower H₂ evolution rates. When an increasing quantity of methyldiazene is added to WT nitrogenase under proton reduction conditions, a progressive inhibition of the H₂ evolution rate is observed (Fig. 2). Methyldiazene was found to be an even stronger inhibitor of proton reduction catalyzed by the α-195^Gln-substituted MoFe protein, providing support for the idea that restriction of proton delivery results in a stronger association of methyldiazene, or one of its reduction products, with the substituted MoFe protein.

Fig. 2.

Methyldiazene inhibition of proton reduction activity. The percentage of maximal H₂ evolution activity (nmol H₂/min per mg of MoFe protein) remaining in the presence of increasing methyldiazene is shown for the WT (△) and α-195^Gln (○) MoFe proteins. The maximum quantity of methyldiazene added to a 9-ml vial assuming 100% yield is shown.

This result suggested it might be possible to trap an intermediate state during turnover of methyldiazene. The addition of methyldiazene to the resting-state α-195^Gln-substituted MoFe protein does not perturb the distinctive resting-state EPR signal of S = 3/2 FeMo cofactor, with g = [4.37, 3.56, 2.01] (Fig. 3). When the α-195^Gln-substituted MoFe protein is freeze-trapped during turnover without added substrate under argon, the FeMo cofactor EPR spectrum is greatly decreased in intensity (31), indicating reduction to an EPR-silent state (data not shown). When methyldiazene was included in the turnover mixture and the sample was freeze-quenched, an EPR spectrum was observed corresponding to an S = ½ spin state. At X band the spectrum appears roughly axial, with g_‖ = 2.08, g_⊥ = 2.02 (Fig. 3); spectra collected at 35 GHz (data not shown) revealed a rhombic signal with g(m) = [2.083, 2.021, 1.993]. This spectrum was not observed when methyldiazene was allowed to decompose before addition to nitrogenase. The methyldiazene-dependent spectrum is similar but not identical to the S = ½ signals of states trapped with other substrates, such as N₂ (26), hydrazine (25), propargyl alcohol (12), acetylene (13), and carbon disulfide (32). Like these other states, this state represents a majority species: conversion of the resting-state FeMo cofactor to this state is over >50% based on spin integration of the EPR signal. Because the α-195^Gln-substituted MoFe protein is compromised in proton delivery (27), the trapped species most likely represents either bound methyldiazene or an early stage in the reduction of methyldiazene. A methyldiazene-dependent EPR signal was not detected when the WT MoFe protein was used. Such binding would be analogous to a middle stage in N₂ reduction, which leads us to use the shorthand m(CH₃NNH) to designate the state. It should be emphasized that whether or not m(CH₃NNH) represents a faithful mimic of an enzyme-bound semireduced N₂ species is not yet established, although our experimental strategy and interpretations are framed around this possibility.

Fig. 3.

EPR spectra are shown for the resting state (Top) and turnover trapped states with methylhydrazine (Middle) or methyldiazene (Bottom) present. EPR conditions are described in Materials and Methods. Calculated g values are noted for some inflections.

To test the possibility that m(CH₃NNH) represents a nonenzymatic breakdown product of methyldiazene, the α-195^Gln MoFe protein was freeze-trapped during turnover in the presence of ammonia, methyl amine, or N₂. These potential breakdown products did not generate an EPR spectrum like that observed when methyldiazene was used as substrate. Nitrogenase trapped during turnover in the presence of another possible breakdown product, methylhydrazine, did reveal a minor EPR signal in the g ≈ 2 region, but that signal was not similar to the one elicited by methyldiazene (Fig. 3).

The microwave power dependence of the EPR signal associated with m(CH₃NNH) was also established (see Fig. 6, which is published as supporting information on the PNAS web site). The results reveal the expected difference of the microwave power saturation for an S = ½ state from that of the S = 3/2 resting-state FeMo cofactor. The microwave power dependence for the m(CH₃NNH)-associated EPR signal is significantly different from that reported for the trapped state formed during hydrazine turnover (25). This latter species is denoted l(N₂H₄) because it is suggested to represent a late stage in N₂ reduction. The pH dependence of the EPR signal intensity for m(CH₃NNH) is also distinctly different from the pH dependence of the EPR signal intensity for the l(N₂H₄) trapped state (see Fig. 7, which is published as supporting information on the PNAS web site). Such features that distinguish m(CH₃NNH) and l(N₂H₄) provides further evidence that the two states are different.

ENDOR Studies of m(HNNCH₃).

To further characterize m(HNNCH₃) and any species bound to FeMo cofactor, methyldiazene was synthesized with ¹⁵N in either the terminal position or the internal position (Scheme 1). As expected, the EPR signal of m(H¹⁵NNCH₃) was indistinguishable from that observed when ¹⁴N-methyldiazene was used. It was shown (26) that the ¹⁵N Mims pulsed ENDOR spectra collected at g₁ = 2.08 for m(¹⁵NHNCH₃) contains a single ¹⁵N doublet that is centered at the ¹⁵N Larmor frequency and has a hyperfine splitting, A(g₁) = 1.5 MHz. In a g₁ spectrum, such a ¹⁵N doublet represents a single type of ¹⁵N of the nitrogenous moiety bound to the FeMo cofactor. A spectrum taken at g₂ = 2.03 (Fig. 4) also exhibits a single doublet, although at this g value the individual peaks have structure and breadth associated with hyperfine anisotropy. These signals are absent in spectra of samples prepared with ¹⁴N-labeled methyldiazene, which demonstrates that m(HNNCH₃) contains a species bound to FeMo cofactor through a nitrogen that originates as the terminal N of methyldiazene.

Fig. 4.

35-GHz ENDOR spectra. (A) ¹⁵N Mims ENDOR spectra of m(CH₃NNH). For H¹⁵NNCH₃, ν = 34.808 GHz, g = 2.03, π/2 = 52 ns, τ = 400 ns, T = 25.352 μs, radio frequency (RF) = 20 μs, eight scans, 30 shots per point, 20-ms shot repetition time, 2 K. For HN¹⁵NCH₃, ν = 34.784 GHz, and other conditions are the same as for H¹⁵NNCH₃. Spectra are normalized for comparison. (B) As in A, but narrower scan and g = 2.01, τ = 552 ns, and 15 scans. Backgrounds have been corrected by simple subtractions as needed. (C) Continuous wave (CW) ¹H-ENDOR of m(CH₃NNH) in H₂O and D₂O. Conditions were: microwave frequency of 35.057–35.171 GHz, modulation amplitude = 4 G, RF sweep speed = 1 MHz/s, bandwidth of RF broadened to 100 kHz, 2 K. Methyldiazene is abbreviated as HNNMe.

The ¹⁵N-ENDOR spectrum associated with m(HN¹⁵NCH₃) at g = 2.03 (Fig. 4) shows a background feature presumed to be from enzyme ¹⁴N, but the ¹⁵N doublet seen with the m(H¹⁵NNCH₃) state is gone; neither is an ¹⁵N doublet seen in broader scans at this and other g values. Narrowing the scan at this and other g values shows only weak features that persist in the natural-abundance sample; for example, a narrow g = 2.01 spectrum shows a single weak feature in this range in both ¹⁵N- and ¹⁴N-labeled trapped states (Fig. 4). Thus, whereas a signal from what originates as the terminal ¹⁵N is readily observed, no signal associated with the ¹⁵N of HN¹⁵NCH₃ is detected. HNN¹³CH₃ and HNNCD₃ also were prepared (4 in Scheme 1) and used in freeze-quench experiments. No differences were observed between the ¹³C or ²H ENDOR spectra of the trapped states when labeled or unlabeled methyldiazene were used as substrate (data not shown). These measurements indicate that the central N from methyldiazene is not bound to a metal ion of FeMo cofactor, or at least not to one with a significant electron-spin projection. In part because of interference from the background ¹⁴N signals, determination of whether the methyldiazene framework is intact or not will require further study and comparison to data from model compounds.

¹H ENDOR spectra of the m(NHNCH₃) state in H₂O and D₂O buffers taken at g₁ = 2.08 (Fig. 4) show an unresolved, nonexchangeable (unchanged in D₂O buffer) central peak at the proton Larmor frequency primarily from the “matrix” protons of nearby residues. As indicated in Fig. 4, the spectra further show a signal from an exchangeable proton(s) with A(g₁) ≈9 MHz. This coupling is too small to be associated with the H⁺/H⁻ bound to FeMo cofactor studied previously (17). It is similar to those of the protons of an alkene bound to FeMo cofactor during alkyne reduction (12–14). As a result, this proton can be assigned to an [-NH_x] moiety that is bound directly to the FeMo cofactor. This assignment of the proton signal, rather than to an exchangeable proton associated with an H-bonding amino acid residue, is supported by comparison with the state that forms upon turnover with N₂ itself (26). That state also displays a ¹⁵N ENDOR signal when formed with ¹⁵N₂ but does not exhibit a resolved signal from an exchangeable proton. With this assignment, m(NHNCH₃) cannot contain bound, deprotonated methyldiazene or the bound [NNHCH₃] isomer (Fig. 5). ¹H couplings to a distal [-NH], if present, would be too small to resolve.

Fig. 5.

Alternating and distal N₂ reduction mechanisms at a metal center (M). Essential intermediates are shown for N₂ reduction mechanisms with distal proton addition (d-mechanism) (Left) and alternating proton addition (a-mechanism) (Right), bound end-on to a metal center (M). Small straight arrows represent the addition of H⁺/e⁻ to substrate. Methyldiazene is placed where it might enter into each mechanism. Only the boxed intermediates are consistent with the ENDOR data for the methyldiazene-derived species trapped on the FeMo cofactor.

Discussion

The biochemical, EPR, ^14,15N-, ¹³C-, and ¹H-ENDOR studies reported here have established: (i) An S = ½ state denoted m(CH₃NNH) is trapped during the reduction of methyldiazene by the α-195^Gln-substituted MoFe protein. (ii) m(CH₃NNH) contains a methyldiazene-derived species bound to the FeMo cofactor. (iii) This species binds through an [-NH_x] fragment whose N derives from the methyldiazene terminal N. (iv) Interaction with the internal N (and methyl) is weak or absent, suggesting that the internal N does not bind to FeMo cofactor. These results provide powerful constraints on possible mechanisms for reduction of methyldiazene by the α-195^Gln MoFe protein and allow steps toward experimental discrimination between the two principal mechanistic views of N₂ reduction.

In the Chatt-Schrock mechanism (22) for N₂ reduction by Mo complexes, H atoms are sequentially added to the distal N of N₂ before N-N bond cleavage. The metal-bound nitride formed by N-N bond cleavage is subsequently reduced to yield the second NH₃ (Fig. 5). This mechanism is denoted as the d-mechanism to indicate the exclusive protonation of the distal N before N-N bond cleavage. For the d-mechanism to operate in the reduction of methyldiazene, the [-NH_x] fragment observed to be bound to FeMo cofactor could be formed only after reduction by ≥3 electrons and only after N-N bond cleavage. Thus, the appearance of bound [-NH_x] is compatible with the d-mechanism only if m(CH₃NNH) represents a late stage in the reduction process. In contrast are mechanisms where H atoms are added alternately to the distal and proximal N before N-N cleavage (denoted as the a-mechanism in Fig. 5) (33–37). An [-NH_x] fragment would be bound at every stage of methyldiazene reduction by an a-mechanism (Fig. 5). These results for m(CH₃NNH) do not rule out the d-mechanism, but make the a-mechanism more attractive because the α-195^Gln-substituted MoFe protein is specifically compromised in its capacity for delivery of protons to nitrogenase substrates, making it unlikely that [-NH_x] could represent a species that is trapped at a late stage of diazene reduction. The findings that N₂H₄ is both a substrate (24, 25) and minor product (23) during N₂ reduction, and that an intermediate can be trapped during N₂H₄ reduction, lend powerful support to the a-mechanism: at no stage in the d-mechanism does N₂H₄ appear.

Additional ENDOR/electron spin-echo envelope modulation studies should disclose more detail about the m(CH₃NNH) state and thus tighten the above constraints. In particular, while the a- and d-mechanisms involve diazene isomers when N₂ is doubly reduced, they incorporate different bound species at reduction levels three and four: the d-mechanism has bound N and NH, whereas the a-mechanism has bound hydrazide and hydrazine species. We anticipate that ENDOR studies of the intermediates trapped with nitrogenous substrates will be able to distinguish among these species. As another ultimately testable difference between the a- and d-mechanisms, each predicts that the first NH₃ is released at different stages of substrate reduction (Fig. 5). Provided that m(CH₃NNH) indeed is a state before N-N bond cleavage, then the present results suggest a fundamental mechanistic difference between reduction of nitrogenous and alkyne substrates. The alkyne substrate-derived intermediates are bound side-on (for two intermediates) (13, 15), whereas the nitrogenous substrate methyldiazene is bound end-on. In any case, mechanistic difference between alkyne and nitrogenous substrates is underscored by the differential effects on substrate reduction elicited by the α-195^Gln substitution within the MoFe protein, where alkyne reduction is unperturbed by this amino acid substitution (27).

Materials and Methods

Materials and Proteins.

All reagents were obtained from Sigma-Aldrich and used as supplied unless stated otherwise. ¹⁵N-labeled hydroxylamine and ¹⁵N-labeled nitromethane were obtained from Cambridge Isotope Laboratories (Andover, MA), and ¹³C- and ²H-labeled nitromethane were obtained from Sigma-Aldrich. Azotobacter vinelandii strains DJ995 (WT MoFe protein), DJ1310 (α-70^Ala MoFe protein), DJ997 (α-195^Gln MoFe protein), and DJ1316 (α-195^Gln/α-70^Ala MoFe protein) were constructed and nitrogenase proteins were expressed as described (25). The MoFe protein from each strain was purified by using a metal affinity chromatography system (38). All proteins used were >95% pure as judged by SDS/PAGE analysis using Coomassie blue staining. Manipulation of proteins was done in septum-sealed serum vials under an argon atmosphere. All transfer of gases and liquids was done with gas-tight syringes.

N-Methylhydroxylamine Synthesis.

The synthesis of isotopically labeled N-methylhydroxylamine (5 in Scheme 1) was accomplished following a modification of a procedure from Ackermann et al. (29) by reduction of nitromethane (4 in Scheme 1) using palladium at room temperature to form the oxalate salt. Briefly, 20 mg of palladium catalyst (5% on barium sulfate) was added to a 50-ml Erlenmeyer flask. To this, 80 mg of oxalic acid and 1 ml of distilled water were added. The flask was then capped and connected with tubing to a second sealed flask (500-ml minimum volume). This second flask was in turn connected with a second piece of tubing to a 500-ml graduated cylinder filled with water. The entire system was flushed with hydrogen gas delivered into the first (50 ml) flask for 5 min, resulting in the catalyst turning black. Finally, 110 μl of ¹⁴N- or ¹⁵N-, ¹³C-, or ²H-labeled nitromethane (4 in Scheme 1) was added via a gas-tight syringe to the first flask containing the catalyst. As hydrogen gas was consumed during the reaction, water from the graduated cylinder was drawn into the second flask, maintaining 1 atm (1 atm = 101.3 kPa) of hydrogen in the first flask. The reaction was run ≈12 h (until hydrogen was no longer consumed). The reaction solution was filtered through a 0.2-μm syringe filter to remove any remaining catalyst, and the product N-methylhydroxylamine (5 in Scheme 1) was recrystallized in absolute ethanol, yielding the oxalate salt upon drying under vacuum. This salt could be stored for months at room temperature in the presence of a desiccant.

Hydroxylamine-O-Sulfonic Acid Synthesis.

The synthesis of hydroxylamine-O-sulfonic acid (3 in Scheme 1) was accomplished by a variation of a published protocol (26, 39). ¹⁴N- or ¹⁵N-labeled hydroxylamine (50 mg; 2 in Scheme 1) was added to a 50-ml Erlenmeyer flask that was then capped with a rubber septum connected to a condenser tube. Methylene chloride (5 ml) was added through the condenser tube, and the reaction was brought to 50°C. Chlorosulfonic acid (100 μl; 1 in Scheme 1) was added to the reaction, and the mixture was allowed to react at 50°C for ≈5 min, resulting in a white flaky crystalline precipitate. The flask was then returned to 25°C, and the methylene chloride solvent was removed with a syringe, leaving the white precipitate. The precipitate was then washed twice with methylene chloride at room temperature (≈2 ml each wash), followed by several washes with diethyl ether at room temperature (≈2 ml each wash) to remove excess acid and water. The resulting white crystals were dried under vacuum for several minutes and used immediately in the synthesis of methyldiazene as described below.

Methyldiazene Synthesis.

Methyldiazene (6 in Scheme 1) was synthesized by using a modification of a method described (40). In a sealed flask (115 ml) under vacuum, 60 mg of hydroxylamine-O-sulfonic acid (3 in Scheme 1) was dissolved in 1.0 ml of water. To this was added a solution (70 mg dissolved in 1.0 ml of water) of N-methylhydroxylamine hydrochloride (Sigma) or the oxalate salt synthesized as described above (5 in Scheme 1) followed by the addition of 1.0 ml of 5 M NaOH. Methyldiazene was evolved as a gas from the solution over the course of several minutes and isolated by freezing in a custom-made gas trap (with a gas sampling port) immersed in liquid nitrogen. The trap containing the frozen methyldiazene was isolated from the rest of the system and quickly warmed to room temperature. Argon was then added to achieve 1 atm pressure. Aliquots of this methyldiazene/argon mixture were withdrawn by syringe and used immediately.

Nitrogenase Activity Assays.

Substrate reduction reactions for nitrogenase proteins were determined as described at pH 7.2 for 10 min at 30°C (41). MoFe protein was added (100 μg) followed by Fe protein (500 μg) to initiate the reaction. Reactions were quenched by the addition of 300 μl of 400 mM EDTA. Hydrogen was quantified by gas chromatography (41). Methyldiazene inhibition of proton reduction was determined as described above except that the pH was adjusted to 7.0.

X-Band EPR Sample Preparation and Analysis.

EPR samples were prepared in a solution containing a MgATP regeneration system as described (25). MoFe protein was added to a final concentration of ≈75 μM (20 mg/ml). Methyldiazene was generated in the same vial containing the protein to maximize the concentration of the substrate. Turnover conditions were initiated by the addition of 50 μM Fe protein. EPR samples under resting conditions were prepared as described above, except that Fe protein was not included. All X-band EPR samples were flash-frozen in liquid nitrogen in 4-mm calibrated quartz EPR tubes. For the pH profile, the buffer was composed of 50 mM Mes, 50 mM 3-[Tris(hydroxymethl)methylamino]-1-propane sulfonic acid, and 50 mM Mops, and the pH was adjusted by the addition of HCl or NaOH. X-band EPR spectra were recorded with a Bruker ESP-300 E spectrometer with an ER 4116 dual-mode X-band cavity equipped with an Oxford Instruments ESR-900 helium flow cryostat. Spectra were obtained at a microwave frequency of ≈9.65 GHz. Initial spectra were obtained at a microwave power of 1.0 mW, a modulation frequency of 1.26 mT, and a temperature of 8 K and were the sum of five scans. The microwave power dependence on EPR signal intensity was determined at 4.7 K with microwave powers ranging from 5 μW to 1 mW. Spin integration of EPR signals was accomplished by comparison to the integrated area of the resting-state FeMo cofactor signal in the MoFe protein.

35-GHz EPR/ENDOR Spectroscopy.

Q-band samples were prepared as described above, except that the MoFe protein concentration was brought to ≈200 μM, and the reactions were initiated with a 100 μM concentration of Fe protein. Continuous wave and Mims pulsed 35-GHz ENDOR spectra were taken as described (15).

Abbreviations

ENDOR

electron nuclear double resonance

N₂

dinitrogen