Evidence for a Two-Electron Transfer Using the All-Ferrous Fe Protein during Nitrogenase Catalysis*

The nitrogenase-catalyzed H2 evolution and acetylene-reduction reactions using Ti(III) and dithionite (DT) as reductants were examined and compared under a variety of conditions. Ti(III) is known to make the all-ferrous Fe protein ([Fe4S4]0) and lowers the amount of ATP hydrolyzed during nitrogenase catalysis by approximately 2-fold. Here we further investigate this behavior and present results consistent with the Fe protein in the [Fe4S4]0 redox state transferring two electrons ([Fe4S4]2+/[Fe4S4]0) per MoFe protein interaction using Ti(III) but transferring only one electron ([Fe4S4]2+/[Fe4S4]1+) using DT. MoFe protein specific activity was measured as a function of Fe:MoFe protein ratio for both a one- and a two-electron transfer reaction, and nearly identical curves were obtained. However, Fe protein specific activity curves as a function of MoFe:Fe protein ratio showed two distinct reactivity patterns. With DT as reductant, typical MoFe inhibition curves were obtained for operation of the [Fe4S4]2+/[Fe4S4]1+redox couple, but with Ti(III) as reductant the [Fe4S4]2+/[Fe4S4]0redox couple was functional and MoFe inhibition was not observed at high MoFe:Fe protein ratios. With Ti(III) as reductant, nitrogenase catalysis produced hyperbolic curves, yielding aV max for the Fe protein specific activity of about 3200 nmol of H2 min−1 mg−1Fe protein, significantly higher than for reactions conducted with DT as reductant. Lag phase experiments (Hageman, R. V., and Burris, R. H. (1978) Proc. Natl. Acad. Sci. U. S. A. 75, 2699–2702) were carried out at MoFe:Fe protein ratios of 100 and 300 using both DT and Ti(III). A lag phase was observed for DT but, with Ti(III) product formation, began immediately and remained linear for over 30 min. Activity measurements using Av-Cp heterologous crosses were examined using both DT and Ti(III) as reductants to compare the reactivity of the [Fe4S4]2+/[Fe4S4]1+and [Fe4S4]2+/[Fe4S4]0redox couples and both were inactive. The results are discussed in terms of the Fe protein transferring two electrons per MoFe protein encounter using the [Fe4S4]2+/[Fe4S4]0redox couple with Ti(III) as reductant.

Nitrogenase is a two-component metalloprotein system that carries out the reduction of dinitrogen to ammonia under mild conditions (1 atm, 30°C) and can also catalyze the reduction of a number of other reducible substrates such as H ϩ , acetylene, etc. (1). The generally accepted view of nitrogenase catalysis with dithionite (DT) 1 as reductant is that the Fe protein (␥ 2 , M r ϳ 63,000) transfers one electron to the MoFe protein (␣ 2 ␤ 2 , M r ϳ 230,000) using the [Fe 4 S 4 ] 2ϩ /[Fe 4 S 4 ] 1ϩ redox couple with concomitant hydrolysis of two ATPs per Fe⅐MoFe protein interaction (1)(2)(3)(4). Using this redox couple, four to five ATPs are hydrolyzed for each pair of electrons transferred to the MoFe protein (5,6). When sufficient electrons have accumulated on the MoFe protein, the bound substrate (N 2 , H ϩ , acetylene, etc.) is reduced to product at the FeMoco cofactor center located within the ␣ subunit of the MoFe protein. Eight electrons are typically required to reduce dinitrogen and two protons to ammonia and H 2 (N 2 ϩ 8H ϩ ϩ 8e ϭ 2NH 3 ϩ H 2 ) accompanied by the hydrolysis of 16 -20 ATPs. Nitrogenase also reduces a variety of other non-physiological substrates using DT as reductant with similar ATP requirements.
With the discovery and characterization (7)(8)(9)(10)(11) of the allferrous Fe protein ([Fe 4 S 4 ] 0 ) and identification of conditions that allow operation of the [Fe 4 S 4 ] 2ϩ /[Fe 4 S 4 ] 0 redox couple to function during catalysis (10,11), new nitrogenase catalytic reactivity has been observed. For example, using 5-7 mM Ti(III) as reductant, nitrogenase is active and the ATP utilization per electron pair transferred (ATP/2e) decreases from 4 -5, typically observed with DT, to values near 2 (10,11). This behavior was proposed to arise from operation of the [Fe 4 S 4 ] 2ϩ /[Fe 4 S 4 ] 0 redox couple with transfer of two electrons to the MoFe protein accompanied by hydrolysis of only two ATPs. Utilization of about one-half as much ATP per Fe protein is the impetus for studying how [Fe 4 S 4 ] 0 interacts with and transfers two electrons to the MoFe protein yet requires hydrolysis of only two molecules of ATP per Fe protein cycle. To better understand this two-electron transfer process and gain insights into the role of the MoFe protein in accommodating both one-and two-electron transfer reactions and coupling them to substrate reduction, we examined several aspects of nitrogenase catalysis involving use of both the [Fe 4  Such characteristic reactions as component protein titrations (6,12), MoFe inhibition (6,12), the dilution effect (13)(14)(15)(16), the delay in product formation (the lag phase) at high MoFe:Fe protein ratios (17,18), and inhibition due to Av-Cp heterologous cross reactions (19 -21)

MATERIALS AND METHODS
All reactions involving the storage and handling of air-sensitive solids or solutions were conducted in a Vacuum Atmospheres glove box under N 2 (O 2 Ͻ 0.1 ppm). TiCl 3 was purchased from Aldrich as a 10% TiCl 3 solution in concentrated HCl. Water and HCl were removed to provide solid anhydrous TiCl 3 (evap) by allowing the solution to stand in a vacuum desiccator over anhydrous CaCl 2 and NaOH while inside a vacuum atmospheres glove box (11). Stock 83 mM Ti(III) in 0.20 M citrate and 0.30 M Tris, pH 8, solutions were prepared in the glove box, standardized optically (⑀ 340 ϭ 730 M Ϫ1 cm Ϫ1 ), and used for the experiments described below (11).
Azotobacter vinelandii MoFe and Fe proteins with specific activities of 1800 -2000 nmol of H 2 min Ϫ1 mg Ϫ1 component protein, determined from dithionite assays, were prepared and characterized as described previously (22). Rick Baer at Exxon Research and Engineering Corporation prepared Cp component proteins with comparable activities. The L127⌬ mutant was a gift from Lance Seefeldt at Utah State University and was prepared as previously reported (23). Standard nitrogenase assays were conducted using both dithionite and Ti(III) as reductants in 5.0-or 12-ml vials under the conditions outlined below using limiting ATP and/or the ATP regeneration system under argon or 10% acetylene in argon. The nitrogenase-catalyzed reaction was quenched after 1-10 min by addition of 0.1 ml of 0.6 M trichloroacetic acid followed by analysis of the gaseous products by gas chromatography. When the regeneration system was used, ATP hydrolysis was measured as inorganic phosphate by the Fiske and Subbarow method (24). For reactions using limiting ATP, nucleotides were measured using an high pressure liquid chromatography method reported previously (10).
Lag phase experiments using either Ti(III) or DT were conducted in 12-ml assay vials. The reaction was started by adding Fe protein to the reaction mixture, and 0.2 ml of the gas was removed from the head space and injected into a Hewlett-Packard 5890 series II gas chromatograph every 2 min. No more than six samples were removed from each reaction vial and corrections were made for the change in volume of the headspace gas due to the removal of consecutive 0.2-ml aliquots. Dilution experiments were carried out as described previously using 20 mM DT or 5.0 -7.0 mM Ti(III) as reductant (25).
Heterologous cross-reaction activity assays were performed using the same techniques as reported above. Assays were run using 1:1 Fe:MoFe protein ratios, a generating system, and excess reductant. Controls were run using the homologous proteins, and H 2 formation was measured by gas chromatography. Additional assays to determine heterologous cross complex formation were performed containing MgATP (no generating system), however the assays were not acid-quenched. Samples from these assays were run on 7% non-stacking native polyacrylamide gels. Polyacrylamide is not readily O 2 -permeable as verified by addition of the redox-active dye bromphenol blue, which is colorless when reduced. Bromphenol blue is readily oxidized by O 2 , but no blue color was observed during electrophoresis, demonstrating that the proteins remained anaerobic. Large protein molecular weight standards were run along side the assay mixture to monitor protein progress and indicate molecular weight size of the bands. Fig. 1 shows the MoFe specific activity curves for H 2 evolution as a function of the Fe:MoFe protein ratio using either 20 mM DT or 7 mM Ti(III) as reductant. At 7.0 mM Ti(III), inhibition by this reductant is minimal, but the [Fe 4 S 4 ] 2ϩ /[Fe 4 S 4 ] 0 redox couple is fully operative (11). Under both sets of conditions, the MoFe specific activity curves were fitted by non-linear least squares and gave K m values for MoFe specific activity of 6.0 Ϯ 0.2 and 6.2 Ϯ 1.2 for DT and Ti(III), respectively. The V max values are quite comparable at 568 Ϯ 10 and 582 Ϯ 37 M min Ϫ1 (2240 Ϯ 40 and 2520 Ϯ 160 nmol of H 2 min Ϫ1 mg Ϫ1 MoFe protein) for DT and Ti(III), respectively. Although similar V max values are obtained that suggest nearly identical behavior, the use of DT and Ti(III) produced distinct ATP utilization patterns characterized by ATP/2e values near 4 and 2, respectively, consistent with previous results (10,12). Nearly identical behavior was seen for acetylene reduction (data not shown). For the acetylene reduction reaction using either DT or Ti(III), about 10% of the total electron flow is directed toward H 2 evolution even though the amount of ATP hydrolyzed is 2-fold higher using DT.

MoFe Protein Titration with Fe Protein-
Fe Protein Titration with MoFe Protein- Fig. 2 compares the Fe protein specific activity as a function of the MoFe:Fe protein ratio for H 2 formation using DT and Ti(III) and shows that the Fe protein specific activity was highly dependent on the reductant used. With DT as reductant, typical MoFe inhibition behavior for H 2 formation was observed causing the rate of product formation to decrease with increasing MoFe:Fe protein ratios above 1.0 -2.0. This result was also observed when low Ti(III) concentrations (ϳ 1.0 mM) were used, due to the predominate operation of the [Fe 4 S 4 ] 2ϩ /[Fe 4 S 4 ] 1ϩ redox couple (data not shown). At a MoFe:Fe protein ratio of about 1.0 -2.0 with DT, the Fe protein specific activity for H 2 formation was at a maximum at 1850 nmol of H 2 min Ϫ1 mg Ϫ1 Fe protein but declined to 20% of this value at a MoFe:Fe protein ratio of 15.  The specific activity of the Fe protein in Fig. 2 is about 1850 nmol of H 2 min Ϫ1 mg Ϫ1 Fe protein as determined by a typical DT assay. In contrast, the measured Fe protein specific activity in Fig. 2 at 7.0 mM Ti(III) is 3100 Ϯ 100 nmol of H 2 min Ϫ1 mg Ϫ1 Fe protein, which is significantly higher than that obtained using DT. Fe protein specific activities as high as 4200 nmol of H 2 min Ϫ1 mg Ϫ1 Fe protein have been observed in separate experiments performed under 10% acetylene or under argon at a 100:1 MoFe:Fe protein ratio using 7.0 mM Ti(III) (data not shown).
The results in Fig. 2 show that MoFe inhibition did not occur using Ti(III) concentrations that support the formation of  (27). ATP/2e values even at MoFe:Fe protein ratios of 100 remained constant at about 2 using Ti(III) as reductant.
Lag Phase Experiments-Hageman and Burris reported (17) that at a MoFe:Fe protein ratio of 100, with DT as reductant, the rate of product formation is nearly zero for 4.3 min before finally reaching a steady state. In contrast, the rate of ATP hydrolysis is constant from the initiation of the reaction. This experiment has formed a cornerstone in our understanding of nitrogenase function by suggesting the MoFe protein randomly accumulates electrons delivered by the Fe protein for eventual substrate reduction. Fig. 3 shows results of the lag phase reactions with both DT and Ti(III) conducted with a MoFe:Fe protein ratio of 300:1. The rate of H 2 production is shown over a 25-min interval, during which the rate of H 2 production has reached a steady state. Extrapolation of the line backward in time produces a near zero time intercept, indicating that product formation occurs immediately after initiation of nitrogenase catalysis and its rate is constant, even beyond the 25-min period of examination. The small positive y intercept is seen in all of the experiments we have conducted. Control reactions containing Ti(III) and the nitrogenase proteins but no ATP regeneration system were used to correct for background H 2 produced by Ti(III) during the course of the catalytic reaction, but it appears that the amount of H 2 produced during nitrogenase turnover may be slightly larger than these controls. ATP hydrolysis also begins immediately upon initiating the reaction and quickly reaches a steady state. Division of the ATP hydrolysis rate by the rate of product formation gives ATP/2e values of near 2. The Fe protein specific activity using Ti(III) at a 300:1 MoFe:Fe protein ratio was 4100 Ϯ 50 nmol of H 2 min Ϫ1 mg Ϫ1 Fe protein, which is consistent with that found in Fig. 2.
Also shown in Fig. 3 is a comparison reaction under the same conditions with DT as reductant. The lag phase of 1.3 Ϯ 0.1 min is much shorter than the 4.3-min lag phase of Hageman and Burris (17) at a ratio of 100:1 MoFe:Fe protein ratio. To further investigate the reason for this behavior, an identical lag phase experiment was carried out under identical conditions to duplicate the original experiment reported by Hageman and Burris (17). Using a 100:1 MoFe:Fe protein ratio, we were not able to duplicate the 4.3-min lag phase but instead measured only a 0.4-min lag (data not shown) at the same Fe protein, ATP, and DT concentrations reported by Hageman and Burris (17). This difference could be attributable to the low Fe protein specific activity of about 130 nmol of H 2 min Ϫ1 mg Ϫ1 Fe protein obtained from their slope, whereas the Fe protein activity using DT and 300:1 and 100:1 MoFe:Fe protein ratios was about 2000 nmol of H 2 min Ϫ1 mg Ϫ1 Fe protein. The greater the Fe protein specific activity, the faster electron transfer occurs to the MoFe protein, resulting in a shorter lag phase. Because the activity of our protein was significantly greater than that used by Hageman and Burris, it was necessary to increase the ratio to 300:1 MoFe:Fe protein to reliably measure a lag phase of 1.3 min with DT as reductant.
To more closely replicate the results of Hageman and Burris, Fe protein with an activity of about 300 nmol of H 2 min Ϫ1 mg Ϫ1 Fe protein, determined by assays using DT and the ATP regeneration system, was used. Fig. 4 compares the results obtained using this partially active Fe protein with both DT and Ti(III) as reductants. A lag phase of 4.1 Ϯ 0.7 min was obtained using DT at a MoFe:Fe protein ratio of 100:1, a value comparable to that reported by Hageman and Burris. In contrast, no lag phase was observed under the same conditions using Ti(III), indicating that apparently the active portion of partially active Dilution Effect-Thorneley and Lowe (13, 28) studied the dilution effect for nitrogenases from Klebsiella pneumoniae (Kp) and Azotobacter chroococcum (Ac) with DT as reductant and proposed that low protein concentrations prevent complex formation between the nitrogenase component proteins. The dilution effect was also recently examined with both Av and Cp and gave results comparable to both Kp and Ac (25). The dilution effect feature that is relevant to this study is the linear increase in product formation as a function of increasing MoFe protein concentration at a fixed Fe:MoFe protein ratio that extrapolates to a finite x intercept near 0.2-0.4 M. Fig. 5 shows this linear section of the dilution effect and the corresponding intercepts conducted using DT and Ti(III) as a function of protein concentration at an Fe:MoFe protein ratio of 1.0. For these experiments, the slopes of the linear portion and the x intercepts were measured and compared. The slopes (rate of product formation) of both the DT and the Ti(III) dilution curves in Fig. 5 are statistically similar. In addition, positive intercepts on the x axis of 0.26 Ϯ 0.02 and 0.34 Ϯ 0.03 M for Ti(III) and DT were obtained, respectively. Several sets of reactions show some variation in the exact x intercept, but in all cases, the Ti(III) curves consistently had a smaller x axis intercept when compared with the DT reactions as shown in Fig. 5.
Heterologous Av2-Cp1, Cp2-Av1 Reactions-Nitrogenase reactions involving Av and Cp heterologous crosses were conducted to determine if the redox state of Fe protein ([Fe 4 S 4 ] 0 or [Fe 4 S 4 ] 1ϩ ) changed the behavior of heterologous cross reactivity previously reported using DT as reductant (13, 19 -21). Results obtained with DT were identical to those previously reported (19,21) showing that the Av1-Cp2 and Cp1-Av2 combinations were totally inactive at 1:1 ratios. When these same combinations react using Ti(III), both combinations were inactive as well.
To determine if the Av and Cp Fe proteins were interacting in the same manner regardless of their redox state, duplicate heterologous combinations at 1:1 protein mixtures were examined. One reaction contained Ti(III) and the other DT under normal assay conditions. After the reaction was initiated, the proteins were removed for native gel electrophoresis. Control reactions containing Av1⅐Av2 and Cp1⅐Cp2 were also prepared, and samples were removed for native electrophoretic analysis under identical conditions. Using a 1:1 Fe:MoFe protein ratio assured that, if a complex formed, the Fe protein band would disappear and the MoFe protein band would be shifted. Both the Ti(III) and DT assays gave a shifted MoFe protein band near 300 kDa and the Fe protein band disappeared. In contrast, no complex formation was observed using either the Ti(III) or the DT with a Cp1⅐Av2 mixture, which is consistent with previous results (19,21).
In addition to these heterologous cross reactions, reaction between the L127⌬ Fe protein (which mimics the ATP hydrolysis transition state) (23) and Av1 was investigated. Although not a true heterologous cross, this combination forms an inactive, tight complex similar to the heterologous crosses above. No product was measured when the L127⌬ Fe protein was mixed at a 1:1 ratio with native Av1 in the presence of 7.0 mM Ti(III). Native Electrophoretic analysis of this mixture showed that the L127⌬ Fe protein band disappeared and an Av1 protein band was once again shifted up to about 300 kDa, indicative of complex formation. Similar results were obtained for DT. ate during nitrogenase catalysis and how they influence electron transfer and ATP hydrolysis during the interaction of the two component proteins.
The titration of the MoFe protein with Fe protein (Fig. 1)  It is of interest now to consider the behavior produced by these two different reductants shown in Fig. 1 as the Fe:MoFe protein ratio decreases. At Fe:MoFe protein ratios below complete saturation of the MoFe protein, the rate at which the MoFe protein generates product is essentially identical and independent of the redox couple that is operating. Additionally, the dilution data show that the rate of product formation at a 1:1 Fe:MoFe protein ratio or any other protein ratio is comparable for the two redox couples over a wide range of concentrations.
In contrast to the results just discussed, a much different behavior is observed when the Fe protein is titrated with the MoFe protein. The Fe protein specific activity curve using DT as reductant has a peak activity of about 2000 nmol of H 2 min Ϫ1 mg Ϫ1 Fe protein at a MoFe:Fe protein ratio of about 1:1 to 2:1. At MoFe:Fe protein ratios above 2:1, the activity decreases due to MoFe inhibition. MoFe inhibition is a characteristic nitrogenase reaction common to all nitrogenases thus far examined and is manifest as decreasing product formation with increasing MoFe protein concentration at constant Fe protein concentration. A proper explanation has remained elusive, although such reactivity has been used to suggest possible stoichiometric relationships for nitrogenase complex formation during catalysis and to assess the specific activity of the Fe protein (29,30). MoFe inhibition was proposed by Johnson et al. (26) to be due to two Fe proteins interacting cooperatively with one active site of the MoFe protein. If this is the case, significantly different behavior would be expected using Ti(III) as is verified in Fig. 2 Another characteristic nitrogenase reaction studied here is the delay in product formation occurring at high MoFe:Fe protein ratios as initially reported by Hageman and Burris (17). In the original report, Hageman and Burris found that it took approximately 8 min for the product formation to reach steady state. When the linear portion of the curve was extrapolated backward in time, it fitted to approximately a 4.3-min lag phase. This experiment was interpreted as a random transfer of electrons from the Fe protein to the MoFe protein with product formation occurring only when two electrons accumulate in the same catalytic center of the MoFe protein. The difference in reactivity between a completely random one-electron electron transfer reaction using the [Fe 4 S 4 ] 2ϩ /[Fe 4 S 4 ] 1ϩ redox couple and a two-electron electron transfer reaction using the [Fe 4 S 4 ] 2ϩ /[Fe 4 S 4 ] 0 redox couple would be quite different. With the random [Fe 4 S 4 ] 2ϩ /[Fe 4 S 4 ] 1ϩ redox couple, a significant delay in product formation would be expected before a steady-state rate of product formation is observed. However, if a concerted two-electron transfer occurs at the same catalytic site on the MoFe protein, then the delay in product formation should be eliminated and product formation would be observed upon initiation of the reaction. Figs. 3  ] 0 redox couples and suggest that the latter operates by transferring two electrons at a time to the MoFe protein center. The difference between our results and those of Hageman and Burris is the protein used here was significantly more active, and this is the reason our reactions were performed at 300:1 MoFe:Fe protein ratio rather than 100:1.
To demonstrate that protein activity influenced the length of the lag phase for the [Fe 4 S 4 ] 2ϩ /[Fe 4 S 4 ] 1ϩ redox couple, we examined a partially inactive Fe protein (ϳ300 nmol of H 2 min Ϫ1 mg Ϫ1 Fe protein). By doing this we were able to more closely replicate the results of Hageman and Burris. Fig. 4 shows the difference between DT and Ti(III) assays run with the 100:1 MoFe:Fe protein ratio and the low activity Fe protein.
The DT assay yielded a lag phase of 4.1 Ϯ 0.7 min, whereas the Ti(III) assay generated product linearly from time zero. Because little has been reported regarding what happens to an Fe protein when it is inactivated, it is difficult to speculate as to