Azotobacter vinelandii Vanadium Nitrogenase: Formaldehyde is a Product of Catalyzed HCN Reduction and Excess Ammonia Arises Directly from Catalyzed Azide Reduction

Abstract

The nitrogenase-catalyzed reduction of both cyanide and azide results in the production of “excess NH₃”, which is an amount of NH₃ over and above that expected to be formed from the well-recognized reactions. Several suggestions have been made as to the possible sources of “excess NH₃”, but previous attempts to characterize these reactions have met with either limited (or no) success or controversy. Because V-nitrogenase has a propensity to release partially reduced intermediates, e.g., N₂H₄ during N₂ reduction, it was selected to probe the reduction of cyanide and azide. Sensitive assay procedures were developed and employed to monitor the production of either HCHO or CH₃OH (its further two-electron-reduced product) from HCN. Like Mo-nitrogenase, V-nitrogenase suffered electron-flux inhibition by CN⁻ (but was much less sensitive than Mo-nitrogenase) but, unlike Mo-nitrogenase, MgATP hydrolysis was also inhibited by CN⁻. V-nitrogenase also released more of the four-electron-reduced intermediate, CH₃NH₂, than did Mo-nitrogenase. At high NaCN concentrations, V-nitrogenase directed a significant percentage of electron flux into “excess NH₃” and, under these conditions, substantial amounts of HCHO, but no CH₃OH, were detected for the first time. With azide, in contrast to Mo-nitrogenase, both total electron flux and MgATP hydrolysis with V-nitrogenase were inhibited. V-nitrogenase, unlike Mo-nitrogenase, showed no preference between the two-electron reduction to N₂-plus-NH₃ and the six-electron reduction to N₂H₄-plus-NH₃. V-nitrogenase formed more “excess NH₃”, but reduction of the N₂ produced by the two-electron reduction of N₃⁻ was not its source. Rather, it was formed directly by the eight-electron reduction of N₃⁻. Unlike Mo-nitrogenase, CO could not completely eliminate either cyanide or azide reduction by V-nitrogenase. CO did, however, eliminate the inhibition of both electron flux and MgATP hydrolysis by CN⁻, but not that caused by azide. These different responses to CO suggest different sites or modes of interaction for these two substrates with V-nitrogenase.

Azotobacter vinelandii is a strictly aerobic N₂-fixing organism that harbors three related, but genetically distinct, nitrogenases. Which one of these enzymes is operational at any particular time depends on which metal ions are present (1). For more than 50 years, the requirement of molybdenum for nitrogenase was thought to be mandatory (2), even though vanadium was demonstrated early on to be almost as stimulatory as Mo to the growth of some bacteria on N₂ as the sole nitrogen source (3). Despite these and other early experiments, the essentiality of Mo became entrenched. However, it is now clear that Mo is not essential for biological nitrogen fixation. Progress through the 1980's clearly identified Mo-independent nitrogenases (4) and the isolation of a V-containing nitrogenase (V-nitrogenase) from A. chroococcum established that one of the alternative systems was V based (5) as predicted by the early studies. The use of either specific DNA probes or the characteristic ethane production from acetylene reduction has established the distribution of the Mo-independent nitrogenases among bacterial species (1, 6-7). Although all N₂-fixing species studied so far have the Mo-containing nitrogenase (Mo-nitrogenase), it is still unclear why some species have only this single nitrogenase, whereas others have either two or three.

Like Mo-nitrogenase, the V- and Fe-nitrogenases comprise two separately purifiable component proteins. Each of these alternative nitrogenases has a specific homodimeric Fe protein¹ (herein called either VnfH or AnfH, respectively) of about 60 kDa with a single [4Fe-4S] cluster. The second component (the VFe protein or FeFe protein) is of about 200 kDa mass but is an α₂β₂γ₂ hexamer rather than an α₂β₂ tetramer as found for the MoFe protein. These larger component proteins likely contain both types of prosthetic group (the cofactor and P-clusters) found in the MoFe protein (8-10), however, x-ray crystallographic structures are available for only the Mo-nitrogenase and include those of the Fe protein (herein called NifH), the MoFe protein, and 2:1 Fe-protein:MoFe protein complexes (11-12).

All three nitrogenases have the same requirements for catalytic activity, namely, MgATP, a low-potential reductant (usually sodium dithionite in vitro), and anaerobiosis. During the catalytic cycle, the two component proteins combine to form a complex, an electron is passed from the Fe protein (NifH, VnfH or AnfH) to its partner (MoFe, VFe or FeFe protein, respectively), and then the complex dissociates with the hydrolysis of MgATP. The relationship between electron transfer and MgATP hydrolysis is often quantified as the ATP:2e⁻ ratio, which under ideal conditions with Mo-nitrogenase is 4:1. Dissociation of the protein-protein complex is necessary to re-reduce the oxidized Fe protein and recharge it with MgATP, thereby making it competent to participate in another electron-transfer cycle (13). Because no substrate is reduced to product by a single electron, even the simplest substrate requires two such cycles. Furthermore, the component proteins of V-nitrogenase are known to form catalytically competent heterologous enzymes with the complementary proteins of Mo-nitrogenase, indicating a broad similarity in mechanism (2, 6-7).

In addition to N₂, nitrogenases can reduce a number of other small molecules or ions, including acetylene and protons. However, both substrate reduction and inhibitor effects are clearly affected by the substitution of Mo by V. For example: (i) the efficiency of N₂ reduction decreases for V-nitrogenase and accounts for only about 50% of total electron flux under 101 kPa N₂ compared to about 75% for Mo-nitrogenase (14); (ii) the preference for C₂H₂ reduction over H⁺ reduction is much less for V-nitrogenase than for Mo-nitrogenase; (iii) during N₂ reduction only, the amount of MgATP hydrolyzed by V-nitrogenase for each electron pair appearing as product increases by 50% (15); (iv) no specific EPR signals are induced by CO during V-nitrogenase turnover (16); and (v) V-nitrogenase releases significant amounts of 4-electron-reduced products, i.e., N₂H₄ during N₂ reduction (14) and C₂H₆ during C₂H₂ reduction (15), whereas wild-type Mo-nitrogenase releases neither.

Surprisingly, although both cyanide and azide have long held prominent positions among the long list of Mo-nitrogenase substrates (17-18), neither substrate has previously been tested with V-nitrogenase. The products of cyanide (HCN) reduction by Mo-nitrogenase are methane-plus-ammonia (a six-electron process) and methylamine (a four-electron process). Only HCN is a substrate, the CN⁻ also present in NaCN solutions is, in contrast, a potent inhibitor of electron flux through the enzyme. HCN reduction has been suggested to occur as shown in Scheme 1 (19-20).

Scheme 1.

HCN reduction catalyzed by Mo-nitrogenase likely occurs by a series of two-electron/two-proton processes (where [ ] indicates either substrate or intermediate bound to the enzyme).

As Scheme 1 shows, the catalyzed production of both CH₃NH₂ and, under some conditions, “excess NH₃”, i.e., an amount of NH₃ in excess of the CH₄ produced, can be attributed to escape of initially formed intermediates from the active site. The four-electron-reduced intermediate on the reduction pathway to CH₄-plus-NH₃ is methylamine, which is stable towards hydrolysis and would be detectable. However, if the two-electron reduced intermediate, methyleneimine (CH₂=NH), escapes the active site, it would be hydrolyzed to HCHO-plus-NH₃ and so could be the source of the “excess NH₃”. Formaldehyde has never been detected as a product of HCN reduction possibly because: (i) both dithionite (21) and cyanide (22) interact with formaldehyde and so interfere with the Nash assay (23); (ii) insufficient “excess NH₃” is produced by Mo-nitrogenase (although in contrast to our earlier results (20), Li et al. (19) showed significant “excess NH₃” production); and (iii) formaldehyde may suffer further two-electron reduction to methanol. We, therefore, turned to V-nitrogenase to probe the reduction of HCN and the likely production of HCHO because of its propensity to release partially reduced intermediates, e.g., N₂H₄ during N₂ reduction. To do so, we developed and employed sensitive assay procedures to monitor HCN-reduction products in an attempt to identify either the released two-electron-reduced intermediate or the products of its hydrolysis.

A similar situation arises with catalyzed azide reduction (24-26; see Scheme 2, reactions [1]-[3]). Here, “excess NH₃”, an amount over the combined amounts of N₂ and N₂H₄ produced, is also usually observed. At moderate concentrations (ca. 10 mM), azide is a good substrate for Mo-nitrogenase and, like HCN reduction, it is reduced to give multiple products. However, unlike catalyzed cyanide reduction, both HN₃ and N₃⁻ are substrates. Azide ion (N₃⁻) is reduced by two electrons to give N₂-plus-NH₃, whereas hydrazoic acid (HN₃) is reduced by six electrons to N₂H₄-plus-NH₃. A major question is whether the “excess NH₃” is a product of further catalyzed reduction of the N₂ produced (see Scheme 2, reaction [2]) or a product of direct reduction of azide to NH₃ (see Scheme 2, reaction [3]). We have also applied V-nitrogenase to gain insight into this question.

Scheme 2.

Azide (both HN₃ and N₃⁻) reduction catalyzed by Mo-nitrogenase occurs by a six-electron/six proton and a two-electron/two-proton process, respectively, to give N₂, N₂H₄, and NH₃. “Excess NH₃” (see text) has been suggested to occur either indirectly by reduction of N₂ formed in reaction [2] or directly by reaction [3].

Experimental Procedures

Cell growth and protein purification

The growth of CA11.71 wild type A. vinelandii strain containing the V-nitrogenase, nitrogenase derepression, and cell-extract preparation were performed as previously described (20) with the following exceptions. Ultra-pure grade chemicals were utilized, contaminating Mo was removed from the sugar and phosphate before addition to the media (27-28) and 1 μM V₂O₅ was added to the media in place of Na₂MoO₄. Protein purification was exactly as described by Kim et al. (29). The purified VFe protein had a specific activity of 1000 nmol H₂ (min.mg protein)^-1, contained 1.1 mol of V per mol protein, and had a 15:1 Fe:V ratio as determined by inductively coupled plasma emission spectroscopy. Because the Fe:V ratio is as expected for the holo-VFe protein, but only about half of the expected V is present, this preparation is likely contaminated with an equal amount of a non-Fe-containing protein. No Mo (<0.01 mol Mo per mol protein) was detected in this preparation. The V:Fe ratio also indicates that there was no loss of metal-containing subunits during purification as previously reported (30). A limited amount of VnfH (specific activity of 1000 nmol H₂.(min.mg protein)^-1) was obtained from these preparations, so that all data presented here use NifH (specific activity of 2200 nmol H₂.(min.mg protein)^-1) as the electron donor to the VFe protein. Some of the experiments were also performed using VnfH and produced identical results.

Nitrogenase assays

Nitrogenase assays were assembled under argon in 9.5-mL glass serum vials with a butyl rubber stopper held on by an aluminum cap. The reaction mixture (1.0 mL final volume) contained 30 μmol creatine phosphate, 25 μmol HEPES buffer (pH7.4), 20 μmol sodium dithionite, 5 μmol MgCl₂, 2.5 μmol ATP, and 0.125 mg creatine phosphokinase. An anaerobic NaCN stock solution was prepared by flushing solid NaCN in a sealed serum vial prior to the addition of degassed 25 mM HEPES (pH7.4) and a predetermined amount of degassed 12 M HCl. This solution was added by syringe to each anaerobic assay vial at the start of the temperature preincubation period. Assays (ca. 0.25 mg total protein in 1 mL unless otherwise stated) were initiated either by the syringe addition of VFe protein, followed immediately by Fe protein (either NifH mostly or VnfH) or by a pre-mixed 1:20 combination of the two. Unless otherwise stated, assays were terminated after 30 min by injection of 0.25 mL 0.5 M EDTA-Na₂ (pH7.5). The time course assays contained 5-times as much of both proteins in 1.5 mL total volume. When required, CO was added by gas-tight Hamilton syringe (Hamilton Syringe Co., Reno, NV) to the appropriate assay vial during the preincubation period.

To determine the effect of H₂ on N₂ formation from N₃⁻ and its subsequent possible reduction to NH₃, 1.5-mL assays containing 10 mM sodium azide, 0.036 mg VFe protein, and a 20-fold molar excess of NifH were run for 40 mins in either the presence or absence of 101 kPa H₂. Assay vials were initially degassed and filled with argon before replacing the argon with H₂. Ammonia and hydrazine were determined as described below.

The effect of pH on azide reduction was monitored in assays that contained either 1 mM or 5 mM azide. By varying the pH over the range 7.15-6.2, HN₃ concentrations varying from 2.8-24.5 μM and 14-122 μM for the 1 mM and 5 mM azide concentrations, respectively, were produced.

Analytical assays

1. H₂ production was quantified gas chromatographically with a molecular sieve 5A column and a thermal conductivity detector. CH₄ was quantified with a Porapak N column and a flame ionization detector with He as the carrier gas.

2. The measurement of MgATP hydrolysis from assays containing a creatine phosphokinase ATP-regenerating system was performed using the colorimetric method of Ennor (31), after pretreatment of samples as described by Dilworth et al. (32).

3. NH₃ production was determined from the same samples by the indophenol method (33) with the pretreatment of samples as described by Dilworth and Fisher (34).

4. CH₃NH₂ was determined from 4-mL assays performed in 37-mL serum vials, containing 2 mg total protein at a 20:1 Fe protein:VFe protein ratio, exactly as previously described (20).

5. N₂H₄ production was determined by adding a 0.8-mL aliquot of an assay to 1.2 mL of a p-dimethylaminobenzaldehyde solution and measuring the resulting absorbance at 458nm (24).

6. N₂ production from fully labelled potassium azide was measured mass spectrometrically on a model 7070 E-HF mass spectrometer (VG analytical, Manchester, UK). The amount of N₂ produced was calculated from the ³⁰N₂ peak compared to the internal Ar peak.

7. Low levels of formaldehyde were measured colorimetrically with acetylacetone/ammonium acetate as described (23) in V-nitrogenase assays that contained 1.25 mg total protein at a 20:1 Fe protein:VFe protein molar ratio and were assembled under Ar in 9.5-mL glass serum vials with a butyl rubber stopper held on by an aluminum cap. Spiked 1.5-mL assays, containing various HCHO concentrations from 0-100 nmol, received 0.2 mL 1 M HCl before the proteins were added to prevent HCN reduction. Fully active assays were terminated after the appropriate incubation period at 30°C by the addition of 0.2 mL 1 M HCl. After analysis of the CH₄ produced, the rubber seals were removed and 30 μL of 1% starch added. Assay vials were weighed prior to being titrated with 0.2 N I₂ to the starch end point (to oxidize the remaining dithionite). After re-weighing to determine the new volume, 40 μL of 2 M AgNO₃ was added (to remove residual cyanide) and the total assay volume was centrifuged for 5 min at 14,000 rpm in a microfuge. Aliquots (1.9 mL) were combined with 30 μL 7.5 M HCl and 0.5 mL of a 4× Nash reagent solution, which contained 0.8 mL acetylacetone, 1.2 mL glacial acetic acid, and 60 g ammonium acetate in 100 mL, prior to incubation at 38°C for 90 min. Samples were centrifuged again and the A₄₁₄ determined in a 4-cm cell.

8. Attempts to measure methanol production employed an enzymatic assay consisting of alcohol oxidase and peroxidase coupled to the oxidation of the chromogen 2,2′-azino-di-(3-ethyl)-benthiazoline-6-sulfonic acid (ABTS) (35). The MgATP to be used in these assays was placed under constant vacuum overnight to remove contaminating ethanol before use. EDTA-quenched assays that either contained a known amount of methanol (0-50 nmols) or had turned over in the presence of sodium cyanide were passed through 6-cm Dowex AG1-X2 (Cl⁻) anion-exchange Pasteur-pipette columns and then washed through with distilled water (3 × 0.5 mL). A 2-mL aliquot of the eluate was applied to 2.5-cm Dowex-50X8 (H⁺) Pasteur-pipette column and washed through with distilled water (2 × 0.5 mL). 1 M NaOH (0.2 mL) was then added to each sample. Each aliquot (1.5 mL) was prepared for assay by addition of O₂-saturated ABTS/phosphate reagent (1 mL) to give a final assay concentration of 100 mM phosphate (pH7.5) and 2.5 mg ABTS. The assay was initiated by addition of 5 units of peroxidase and 0.01 units alcohol oxidase and the A₄₁₄ determined after both 30-min and 60-min incubation at room temperature.

Results

Product formation as a function of NaCN concentration

The formation of CH₄, H₂ and NH₃ as a function of NaCN concentration is presented in Figure 1 (top). Methylamine was not determined from these assays and no adjustment for this omission has been made. The amount of H₂ produced decreases with increasing NaCN concentration and the amount of CH₄ formed increases initially but then declines at NaCN concentrations greater than 5mM in parallel with H₂ production. In contrast, NH₃ production increases with increasing NaCN concentration up to 20 mM NaCN, but remains constant from 20 mM to 50 mM NaCN. These differing responses to NaCN concentration result in ca. 70 nmols “excess NH₃” being produced at the higher NaCN concentrations. Figure 1 (bottom) shows the corresponding inhibition of electron flux and the ATP:2e⁻ ratio determined from the data presented in Figure 1 (top) (with no accounting of electrons used for methylamine production). These data also show that HCN reduction by V-nitrogenase is much less sensitive to cyanide inhibition (presumably by CN⁻) than is Mo-nitrogenase. Further, unlike Mo-nitrogenase, this inhibition of activity is accompanied by a similar decrease in the rate of ATP hydrolysis, which allows electron transfer to remain coupled to ATP hydrolysis (20).

Figure 1.

(Top). Effect of NaCN concentration on H₂ (O), CH₄ (◊) and NH₃ (Δ) formation by V-nitrogenase. Assays were performed under 101 kPa Ar and each assay contained 0.039 mg of VFe protein and 0.21 mg of Fe protein (NifH). Assays were terminated after 12 mins. (Bottom). Inhibition of total electron flux (calculated as 2e⁻ for H₂, 6e⁻ for CH₄, and 2e⁻ for “excess NH₃”; O) and the resulting ATP:2e⁻ ratio (Δ) as a function of increasing NaCN concentration.

In a separate experiment, the K_m value for CH₄ production from V-nitrogenase catalyzed NaCN reduction over the range 1-3.3 mM was calculated as 1.9 mM. This value corresponds closely to the published value of 1.6 mM for the Mo-nitrogenase (20).

Methylamine determination

The complexity of the method used to determine methylamine formation (20) made it impossible to measure product formation at all NaCN concentrations shown in Figure 1 (top). We, therefore, concentrated on assays containing either 5 or 50 mM NaCN because they produce either none or significant amounts of “excess NH₃”, respectively. Although in the presence of 5 mM NaCN no “excess NH₃” is produced, CH₃NH₂ is clearly detectable. In fact, the CH₃NH₂:CH₄ ratio is 0.66:1, which is much higher than the published value of 0.35-0.39:1 for the Mo-nitrogenase under similar conditions (19-20). The rate of CH₃NH₂ production increases significantly at the higher NaCN concentration to become similar to that for methane production (which decreases), resulting in a CH₃NH₂:CH₄ ratio of ca. 1:1. Together with this increased CH₃NH₂-production rate, there is also an increase in the “excess NH₃”-formation rate (see Table 1).

Table 1.

Product formation, CH₃NH₂:CH₄ ratio, and ATP:2e⁻ ratio from 5 mM and 50 mM NaCN by V-nitrogenase and 5 mM NaCN by Mo-nitrogenase^a.

N2aseb	NaCN (mM)	specific activity [nmol of electron pairs (min.mg protein)−1]				CH3NH2 CH4 ratio	ATP:2e−
		H2	CH4	NH3	CH3NH2
V	0	980±30	0	0	0	--	4.9
V	5	560±9	59±1.5	59±4.5	39±0.3	0.66:1	5.9
Moc	5	270	150	150	58	0.39:1	20
V	50	256±21	50±1.5	65±0.5	54±6	1.1:1	6.8

^aNitrogenase assays were conducted in 4-mL reaction volumes under 101 kPa Ar with a 20-fold molar excess of Fe protein (NifH).

^bN₂ase = nitrogenase.

^cMo-nitrogenase data taken from Fisher et al. (20).

Recovery of HCHO from steady state assays

Before attempting to determine if HCHO was a product of HCN reduction, it was important to ensure that HCHO could be recovered and quantified when added to a typical nitrogenase assay. To do so, known amounts of HCHO were added to: (i) assays turning over under 101 kPa argon in the absence of HCN; and (ii) assays that contained HCN and nitrogenase proteins but which were acidified with HCl so that no catalysis occurred. Figure 2 shows a typical HCHO-recovery curve developed as described in Experimental Procedures for pre-acidified (“killed”) assays containing 50 mM NaCN. When assays were allowed to turnover under an argon atmosphere in the absence of NaCN, the results were essentially identical (data not shown). Recovery was 80% or better.

Figure 2.

HCHO-recovery curve for HCHO added to “pre-killed” nitrogenase assays containing 50mM NaCN and 1.25 mg total protein at a 20:1 NifH:VFe protein molar ratio. Aliquots from vials containing a known amount of HCHO were assayed as described under Experimental Procedures.

Time Course of HCHO formation from HCN reduction by V-nitrogenase

Figure 3 shows the production of HCHO, “excess NH₃”, and CH₄ as a function of time from nitrogenase assays containing 50 mM NaCN. The rates of formation of all three products were effectively linear for the first 15 min, but thereafter all rates decreased until, by 30 min, product formation had effectively ceased. This loss of linearity likely reflects a MgATP and/or dithionite limitation because these assays contained five-times as much protein but only 1.5-times as much MgATP and dithionite as normal assays. Ammonia and CH₄ production were determined from the same assay vials, which were quenched by the addition of Na₂-EDTA, and showed that up to 100 nmol “excess NH₃” had been produced at assay times of 15 min or longer. HCHO production was measured using duplicate assays, which were quenched with HCl, and showed that ca. 50 nmol HCHO had been produced over the same time period.

Figure 3.

Time course for the production of CH₄ (O), HCHO (◊), and “excess NH₃” (Δ) by V-nitrogenase. Assays were identical to those in Figure 2, except no HCHO was added and assays were allowed to run for 5-30 mins. Total NH₃ formed was determined from assays that were quenched by the addition of 0.35 mL 0.5M Na₂-EDTA, but only “excess NH₃” production is plotted.

Attempts to detect methanol as a product of HCN reduction

Methanol (10-30 nmol) added to quenched nitrogenase assays was recovered in 75-80% yield after a 60-min incubation at room temperature with the ABTS/phosphate reagent. But “live” assays, which contained 50 mM NaCN and no added methanol, showed no increase in absorbance at 414 nm when compared to quenched control samples after being allowed to run for 40 min. Therefore, although these assays produced ca. 100 nmol “excess NH₃”, they produced no methanol.

Azide reduction by VFe protein

The apparent K_m for azide was determined at pH 7.4 by measuring the rate of N₂H₄ formation as a function of potassium azide concentration (0.8-20 mM). The value obtained was 2.6 mM (ca. 4 μM HN₃; data not shown), which is similar to the 12 μM HN₃ value previously determined for the Mo-nitrogenase (25).

Table 2 shows that both N₂H₄ and N₂ were products of azide reduction with the V-nitrogenase. Assuming similar reactivity to that of Mo-nitrogenase, then N₃⁻ is reduced to N₂ and HN₃ is reduced to N₂H₄, but at only 8% and 33%, respectively, of the rates with Mo-nitrogenase. Even so, both the VFe protein and the MoFe protein produce similar amounts of “excess NH₃”, i.e., NH₃ produced in excess of the sum of N₂ and N₂H₄, but that produced by V-nitrogenase represents 8% of total electron flux compared to only 3% for Mo-nitrogenase. Further, the VFe protein has equal rates of N₂ and N₂H₄ production, whereas the MoFe protein's rate of N₂ production is four-times that for N₂H₄ formation. The amount of N₂ produced constitutes 7% and 21% of the total electron flux measured from the VFe and MoFe proteins, respectively, whereas total NH₃ formed is a similar percentage (22% vs. 30%, respectively) of total electron flux for both enzymes. Unlike Mo-nitrogenase, V-nitrogenase suffers ca. 35% inhibition of both total electron flux and the rate of MgATP hydrolysis by 10 mM azide.

Table 2.

Product formation and ATP:2e⁻ ratio for azide reduction by V-nitrogenase and Mo-nitrogenase^a.

N2aseb	specific activity [nmol of electron pairs (min.mg protein)−1]				ATP:2e−
	N2	N2H4	NH3	H2
V	51±3	53±1	161±3	470±1	5.6
Moc	630	160	880	1300	5.3

^aNitrogenase assays were conducted in 4-mL reaction volumes under 101 kPa Ar with a 20-fold molar excess of Fe protein (NifH) in the presence of 10 mM K¹⁵N₃.

^bN₂ase = nitrogenase.

^cMo-nitrogenase data taken from Fisher et al. (26).

Is azide or hydrazoic acid the inhibitor of electron flux?

The source of the inhibition of total electron flux induced by added azide was probed by monitoring H₂ evolution as a function of pH at either 1 mM or 5 mM sodium azide. Because HN₃ is a weak acid with a pK_a of 4.6 at 30°C, varying the pH between 6-7 results in the calculated concentrations of HN₃ and N₃⁻ listed in Table 3, which also shows the resulting data for H₂ evolution. The trend is the same in both experiments and shows that the inhibition of H₂ evolution increases with increasing HN₃ concentration. The effect of HN₃ is more obvious with 1 mM azide, where inhibition increases by ca. 3-fold as the HN₃ concentration increases by ca. 9-fold. From these experiments, the rate of N₂H₄ production as a function HN₃ concentration gives a K_m value of 4 μM.

Table 3.

Effect of pH on HN₃ and N₃⁻ concentrations and the resulting inhibition of H₂ evolution by V-nitrogenase. Experiments were conducted as described in Experimental Procedures.

pH	1 mM azide			5 mM azide
	Calculated [HN3] (μM)	Calculated [N3-] (μM)	% Inhibition	Calculated [HN3] (μM)	Calculated [N3-] (μM)	% Inhibition
7.15	2.8	997	11	14	4986	36
6.78	6.6	994	17	33	4967	51
6.50	12.4	988	23	62	4938	50
6.20	24.5	976	32	122	4877	54

Is the N₂ produced during N₃⁻ reduction the source of “excess NH₃”?

Because H₂ is a specific inhibitor of N₂ reduction, parallel experiments with V-nitrogenase were conducted with 10 mM sodium azide under either 101 kPa Ar or 101 kPa H₂. No significant difference was found in either N₂H₄ or NH₃ production. Assays conducted under H₂ yielded 169± 2 nmols NH₃ and 37±0.2 nmols N₂H₄, whereas identical assays performed under an argon atmosphere yielded 161±3 nmols NH₃ and 37±1 nmols N₂H₄.

Effect of CO concentration on sodium azide and cyanide reduction

Figure 4 shows the effect of CO on VFe protein-catalyzed azide reduction. As the CO concentration is increased incrementally to 10%, the amounts of both NH₃ and N₂H₄ decrease but their loss is compensated by increases in the amount of H₂ evolved. However, unlike for Mo-nitrogenase, both NH₃ and N₂H₄ are still produced (at ca. 25% of their amounts under 101 kPa Ar) in the presence of 10% CO. Furthermore, this CO-induced redistribution of electron flux does not relieve the inhibition of total electron flux caused by 10 mM NaN₃, which remained at 28% under 10% CO. There is also no significant change in the ATP:2e⁻ ratio at high CO concentrations, indicating that electron transfer remains coupled to ATP hydrolysis.

Figure 4.

The effect of CO concentration on azide reduction to NH₃ (O) and N₂H₄ (◊), plus concomitant H₂ evolution (Δ), by V-nitrogenase. Assays were performed under 101 kPa Ar and the 1.5-mL reaction contained 10 mM NaN₃, 0.039 mg VFe protein at a 20:1 NifH:VFe molar ratio. Appropriate amounts of CO were added with a gas-tight Hamilton syringe to give the desired final concentration. Assays were terminated after 45 min by the addition of 0.4 mL 0.5 M Na₂-EDTA.

Figure 5 shows that CO has a similar effect to that observed during azide reduction when added to assays containing 10 mM NaCN. Again, the highest CO concentration tested (10%) is unable to completely inhibit HCN reduction. Approximately 20% and 15%, respectively, of the amounts of NH₃ and CH₄ formed in the absence of CO are produced under 10% CO. Electron transfer remains coupled to ATP hydrolysis in both the presence and absence of CO. However, CO relieves the 30% “dead-end” inhibition of total electron flux caused by the 10 mM NaCN.

Figure 5.

The effect of CO concentration on HCN reduction to NH₃ (O) and CH₄ (◊), plus concomitant H₂ evolution (Δ), by V-nitrogenase. Standard 1-mL nitrogenase assays were conducted under 101 kPa Ar and contained 10 mM NaCN, 0.039 mg VFe protein at a 20:1 NifH:VFe molar ratio. Appropriate amounts of CO were added with a gas-tight Hamilton syringe to give the desired final concentration. Assays were terminated after 30 min by the addition of 0.25 mL 0.5 M Na₂-EDTA.

Discussion

The role of the heterometal (Mo or V) in nitrogenase function and, in particular, its potential to either bind substrates or influence their reduction continues to be debated. Although a substantial amount of evidence is accumulating that invokes the central iron atoms at the waist of the FeMo-cofactor of the MoFe protein as a binding site for alkynes and alkenes (36-42), the binding site of other substrates, in particular, the most important substrate, N₂, remains uncertain. Replacing Mo by V gives rise to an enzyme still capable of catalyzing all the typical nitrogenase reactions, albeit usually in a less efficient manner (see Introduction). Heterometal replacement provides an alternative approach to investigate the mechanism of biological nitrogen fixation as do studies of the reduction of alternative substrates. Although neither azide nor cyanide has been tested with V-nitrogenase, they were both expected to be substrates, but likely less efficiently used than with Mo-nitrogenase.

Catalyzed HCN Reduction

In general terms, V-nitrogenase catalyzes the reduction of HCN quite effectively when compared to Mo-nitrogenase. Total electron flux through V-nitrogenase is inhibited by CN⁻ as found for Mo-nitrogenase, however, 50% inhibition requires about 15mM NaCN compared to only 5mM for Mo-nitrogenase. Furthermore, unlike Mo-nitrogenase, the rate of MgATP hydrolysis by V-nitrogenase is also similarly inhibited, resulting in electron transfer and MgATP hydrolysis remaining tightly coupled. This “dead-end” inhibition by CN⁻ is similar to that seen with high concentrations of NaCl (43-44). These observations indicate that the “single HCN/CN⁻ binding site” hypothesis (45), in which CN⁻ binds and acts as an inhibitor until it is protonated to give the substrate, HCN, is unlikely with V-nitrogenase.

This high sensitivity to CN⁻ shown by Mo-nitrogenase presented a significant problem in our attempts to detect HCHO produced from catalyzed HCN reduction. When lower NaCN concentrations (up to 5 mM) were used, very little (or no) “excess NH₃” was produced, indicating that very little (or no) HCHO would be present (see Scheme 1 and Table 1) and our assay detected none. If higher NaCN concentrations were used (10-20 mM), then CN⁻ inhibition of electron flux through Mo-nitrogenase became sufficiently severe that the compromised turnover resulted in only a small amount of “excess NH₃” and putatively insufficient HCHO for detection. However, two important differences between Mo-nitrogenase and V-nitrogenase suggested that the latter might offer a better opportunity for detecting HCHO. First, at higher NaCN concentrations, the VFe protein produces significant amounts of “excess NH₃”, whereas the MoFe protein produces much less. Second, the VFe protein releases considerably more CH₃NH₂ than does the MoFe protein (as judged by the CH₃NH₂/CH₄ ratio; see Table 1), consistent with a “leakier” active site.

Previous attempts to measure HCHO added to Mo-nitrogenase assays in the presence of both NaCN and dithionite had failed and resulted in the suggestion that any HCHO formed might not survive long enough to be detected in an assay (19). Although the HCHO-assay protocol that we developed showed that HCHO added to assays is recoverable, we too failed to detect HCHO as a product of Mo-nitrogenase-catalyzed reduction of HCN for the reasons stated above. After attempting to detect and measure methanol, the likely two-electron-reduced product of further HCHO reduction, also without success, we chose to exploit the abundance of “excess NH₃” and the efficient turnover of V-nitrogenase under high NaCN concentrations and returned to HCHO as the likely product of the reaction that produces “excess NH₃”.

The HCHO assay (23) with which we started was fraught with problems for our purpose. We found that residual dithionite inhibited color development with the Nash reagent. However, even after I₂ oxidation of the dithionite (20), color development remained inconsistent. Only after also removing residual cyanide by adding AgNO₃ were we able to recover greater than 80% of the HCHO added to assays.

Applying this protocol to assays of HCN reduction catalyzed by V-nitrogenase, product HCHO was detected and quantified. However, the amount of HCHO detected was significantly less than the amount of “excess NH₃” produced in the same assay vial. This discrepancy could be due problems, such as low sensitivity and/or quantification, inherent to the assay; to another component of the assay that prevents full color development; or to formation of another product that remains undetected. If the latter, it is unlikely to be HCHO's two-electron reduction product, CH₃OH, for which we searched unsuccessfully.

Catalyzed Azide Reduction

Our studies of V-nitrogenase-catalyzed NaN₃ reduction also produced some interesting insights. Although the rates of formation of both N₂H₄ and N₂ are considerably lower than with Mo-nitrogenase, the K_m for N₂H₄ formation is actually lower than that of Mo-nitrogenase, indicating a higher apparent affinity for HN₃. We have found similar seemingly anomalous situations before (46). V-nitrogenase exhibits no preference for the two-electron reduction of N₃⁻ to N₂-plus-NH₃ over the six-electron reduction of HN₃ to N₂H₄-plus-NH₃, whereas Mo-nitrogenase shows a clear preference for the two-electron process. Furthermore, “excess NH₃” production by V-nitrogenase is significantly higher than for Mo-nitrogenase. All of these product-formation preferences by V-nitrogenase likely reflect its higher apparent affinity for substrate than that of Mo-nitrogenase.

Unlike Mo-nitrogenase (see Table 2; 25-26, 47), added NaN₃ inhibits equally both electron flux and the rate of MgATP hydrolysis with V-nitrogenase, resulting in an unchanging ATP:2e⁻ ratio; another example of “dead-end” inhibition. This difference may again reflect the stronger interaction between all azide forms and V-nitrogenase. The source of this inhibition most likely resides with HN₃ rather than N₃⁻.

The source of “excess NH₃” production from catalyzed azide reduction was investigated using the well documented specific inhibition of N₂ reduction by H₂ (14-15, 47). Parallel azide assays, one set under argon and another under H₂, resulted in identical rates of product formation. These results indicate that “excess NH₃” formation is not a result of the further catalyzed reduction of product N₂ because “excess NH₃” formation would be expected to decrease in the presence of H₂ and it did not. Thus, equation [3] rather than equation [2] in Scheme 2 represents the source of “excess NH₃” for V-nitrogenase. Previously, we reached the same conclusion for both wild-type and some variant Mo-nitrogenases (20, 47). However, there are data in the literature which indicate that added H₂ does inhibit “excess NH₃” formation from azide with wild-type Mo-nitrogenase (19). If this were so, then the mechanisms of azide reduction by Mo-nitrogenase and V-nitrogenase would be quite different.

Effect of Carbon Monoxide on Substrate Reduction

CO is a potent inhibitor of all Mo-nitrogenase-catalyzed substrate reductions, except of that of protons. The situation with V-nitrogenase is less well defined but apparently more complex (15, 30, 36, 48-50). Both azide and cyanide reduction by Mo-nitrogenase are completely eliminated by 1 kPa CO (19-20, 24), however, even at 10 kPa CO, substantial substrate reduction continues with V-nitrogenase. Under these conditions, CO relieves the “dead-end” inhibition by cyanide completely; the resulting increased H₂ evolution compensates for both increased electron flux and decreased HCN reduction. Under all conditions, electron flux remains tightly coupled to MgATP hydrolysis.

The effects of added CO on cyanide reduction by V-nitrogenase do not fully carry over to azide reduction. One similarity is that substrate reduction continues at a diminished rate with added CO. However, unlike with cyanide, this redistribution of electron flux does not relieve the “dead-end” inhibition phenomenon. Both azide-induced electron-flux inhibition and the rate of MgATP hydrolysis are unaffected by added CO, resulting in an unchanging ATP:2e⁻ ratio. These responses to CO are completely different to those of Mo-nitrogenase, where there is neither electron-flux inhibition nor MgATP-hydrolysis inhibition by azide at pH 7.4 and where as little as 1 kPa CO completely eliminates all product formation from azide (47). These results are consistent with weaker binding of CO to V compared to Mo and are compatible with an earlier observation that CO is a less effective inhibitor of C₂H₄ production from C₂H₂ by V-nitrogenase than with Mo-nitrogenase (14).

Although all available evidence indicates that the structures of the FeMo-cofactor and FeV-cofactor are similar, our data show that the presence of V obviously impacts catalyzed substrate reduction. Would such an impact be expected if the heterometal has little (or no) involvement in the electronic structure of the cofactor? And, consequently, on catalysis? These questions remain open.