Flavodoxin hydroquinone reduces Azotobacter vinelandii Fe protein to the all-ferrous redox state with a S = 0 spin state

Abstract

Azotobacter vinelandii flavodoxin hydroquinone (FldHQ) is a physiological reductant to nitrogenase supporting catalysis that is twice as energy efficient (ATP/2e^– = 2) as dithionite (ATP/2e^– = 4). This catalytic efficiency results from reduction of Fe protein from A. vinelandii (Av2) to the all-ferrous oxidation state ([Fe₄S₄]⁰), in contrast to dithionite, which only reduces Av2 to the [Fe₄S₄]¹⁺ state. Like FldHQ, Ti(III) citrate yields ATP/2e^– = 2, and Ti(III)-reduced [Fe₄S₄]⁰ Av2 has a S = 4 spin state and characteristic Mossbauer spectrum, a parallel mode g = 16.4 EPR signal, and a shoulder at 520 nm in its UV-vis spectrum, each of which distinguish the S = 4 [Fe₄S₄]⁰ Av2 from other states. In this study, we demonstrate that FldHQ makes [Fe₄S₄]⁰ Av2, which is sufficiently characterized to demonstrate unique physical properties that distinguish it from the previously characterized Ti(III)-reduced [Fe₄S₄]⁰ Av2. In particular, Evans NMR magnetic susceptibility and EPR measurements indicate that FldHQ-reduced [Fe₄S₄]⁰ Av2 has an S = 0 spin state (like [Fe₄S₄]²⁺ Av2). There is no g = 16.4 EPR signal and no shoulder at 520 nm in its absorbance spectrum, which resembles that of [Fe₄S₄]¹⁺ Av2. That the physiological reductant to Av2 is capable of forming [Fe₄S₄]⁰ Av2 has important implications for in vivo nitrogenase activity.

Nitrogenase is a two-component enzyme system present in a number of diverse bacterial species that is responsible for the biological reduction of nitrogen to ammonia. The Fe protein component is the site of MgAXP binding (MgATP or MgADP) and possesses a [Fe₄S₄] cluster, which is reduced by a low potential source of electrons (1). The catalytically active form of the Fe protein is reduced and bound to two MgATP (2). Upon binding the MoFe protein, electron transfer from the Fe protein to the MoFe protein is accelerated by ATP hydrolysis (3).

The Fe protein from Azotobacter vinelandii, Av2, is thought to transfer one electron to the MoFe protein from A. vinelandii (Av1) by using the [Fe₄S₄]²⁺/[Fe₄S₄]¹⁺ redox couple (2+/1+). The discovery (4) and characterization (5–12) of an [Fe₄S₄]⁰ Av2 has raised questions regarding how nitrogenase functions (13–15) because the 2+/0 couple involves two electrons instead of one (5–7) to decrease the optimal ATP requirement 2-fold. This energy efficiency makes the study of [Fe₄S₄]⁰ Av2 of great interest.

Initial formation of the [Fe₄S₄]⁰ cluster by reduced methyl viologen (MV) reported a spectrum similar to [Fe₄S₄]¹⁺ Av2 but of lower intensity. The perpendicular mode EPR signal of [Fe₄S₄]¹⁺ disappeared during reduction by MV, and an EPR silent [Fe₄S₄]⁰ cluster was formed. A redox potential for the 1+/0 couple of –460 mV and an n value of 1.0 was reported (16).

Subsequent characterization of [Fe₄S₄]⁰ Av2 was conducted with other reductants, most notably Ti(III) citrate (17), and an E_m = –790 mV (18) was estimated for the 1+/0 couple, which is in line with E_m estimates of all-ferrous Fe-S clusters by using discrete Fourier transform calculations (19). Indeed, this latter study raised the question of whether [Fe₄S₄]⁰ Av2 can even be made in vivo (1). A value of –790 mV is not consistent with turnover potential measured by Ti(III) and other reductants and with reported potentials for the Ti(IV)/Ti(III) couple (7, 20). If an E_m = –460 mV for the 1+/0 couple (16) is correct, then several in vivo redox proteins could reduce Av2 to the [Fe₄S₄]⁰ state (18). If the more negative E_m = –790 mV value is correct, then [Fe₄S₄]⁰ Av2 is not likely to be physiologically relevant (1). The question of whether [Fe₄S₄]⁰ Av2 is physiologically relevant may hinge on what is the correct E_m.

Flavodoxin hydroquinone (FldHQ) is a common in vivo reductant to nitrogenase (21–24). Using AvFldHQ (E_m = –515 mV; ref. 25), the ATP/2e^– ratio decreased to 2, suggesting that FldHQ supported the 2+/0 redox couple (5). However, the distinguishing characteristics of [Fe₄S₄]⁰ Av2 made by reduction with Ti(III), including a UV-vis shoulder at 520 nm and a strong g = 16.4 EPR signal (8), are absent with both FldHQ- and MV-reduced [Fe₄S₄]⁰ Av2. This result raised uncertainty regarding whether these reductants actually reduce Av2 to the [Fe₄S₄]⁰ state (18). Such concern based on the lack of physical similarities to Ti(III)-reduced [Fe₄S₄]⁰ Av2 is understandable but neglects the possibility that [Fe₄S₄]⁰ Av2 forms and functions with another spin state. [Fe₄S₄]⁰ Av2 can have spin states of S = 0, 4, and 8 (19). Ti(III)-reduced [Fe₄S₄]⁰ Av2 has a spin of S = 4 (8). Of the other two possibilities, S = 0 [Fe₄S₄]⁰ Av2 is closest in energy to S = 4 [Fe₄S₄]⁰ Av2, whereas the S = 8 spin state requires a higher energy (19). Here, we provide evidence that FldHQ reduces Av2 to the S = 0 [Fe₄S₄]⁰ state, accounting for the absence of the distinguishing characteristics of the Ti(III)-reduced [Fe₄S₄]⁰ Av2 when FldHQ is used as reductant.

The characterization of this previously undescribed S = 0 form of [Fe₄S₄]⁰ Av2 coincides with the ATP/2e^– = 2 previously reported for FldHQ-reduced [Fe₄S₄]⁰ Av2. Because nucleotides are required for catalysis, it is imperative that reduction of nucleotide-bound [Fe₄S₄]¹⁺ Av2(MgAXP)₂ should occur readily enough to yield the observed ATP/2e^– ratio of 2. Independent of catalysis, we show that FldHQ rapidly reduces ≈40% of [Fe₄S₄]¹⁺ Av2(MgADP)₂ and ≈73% of [Fe₄S₄]¹⁺ Av2(MgATP)₂ to the [Fe₄S₄]⁰ state, satisfying the nucleotide requirement for the physiological relevance of [Fe₄S₄]⁰ Av2.

Results and Discussion

FldHQ Reduction of the Fe Protein.

Excess sodium dithionite (DT) at pH 8 reduces flavodoxin in an arbitrary state (Fld) to the FldHQ state (26, 27). Oxidized flavodoxin quinone is a yellow, oxidized species (quinone), radical singly reduced flavodoxin semiquinone (FldSQ) is a blue radical species (semiquinone), and FldHQ is a pale yellow, fully reduced species (hydroquinone) as reported by Hinkson and Bulen (28). The FldHQ/FldSQ couple is relevant to nitrogenase catalysis (21, 24) because it is sufficiently negative (–515 mV) to conduct nitrogenase catalysis (25). The maximum optical change for oxidation of FldHQ to FldSQ occurs at 580 nm (Δε = 5.4 mM^–1·cm^–1; refs. 21 and 28). This optical change was used to follow Av2 reduction by monitoring FldSQ formation.

Fig. 1 shows that rapid production of FldSQ occurs upon addition of either [Fe₄S₄]²⁺ or [Fe₄S₄]¹⁺ Av2 to excess FldHQ. From the amount of Av2 added and the amount of FldSQ formed, the electrons transferred to Av2 were determined. This approach was validated by adding [Fe(CN)₆]^3– to FldHQ, which mirrored the reaction of FldHQ with [Fe₄S₄]¹⁺ Av2 (Fig. 1). Addition of anaerobic buffer did not cause formation of FldSQ. Based on seven independent Av2 additions, [Fe₄S₄]²⁺ Av2 was rapidly reduced by 1.92 ± 0.13 electrons, [Fe₄S₄]¹⁺ Av2 by 0.95 ± 0.08, and Ti(III)-reduced [Fe₄S₄]⁰ Av2 by none. These results show that FldHQ generates [Fe₄S₄]⁰ Av2. The FldHQ/Av2 ratio was varied from 3 to 20, and complete reduction of Av2 occurred even at the lowest ratio of Fld/Av2 tested. This result requires that the E_m for [Fe₄S₄]⁰ Av2 formation is more positive than the FldSQ/FldHQ potential of –515 mV. Thus, this result supports the originally measured value of –460 mV rather than the more recent value of –790 mV for the 1+/0 couple. Possible sources for the discrepancy are as follows: (i) a conformational change in Av2 induced by interaction with reductant that changes its reduction potential and/or (ii) a conformational change in Fld induced by binding to Av2 that changes its reduction potential.

Fig. 1.

Formation of [Fe₄S₄]⁰ Av2 from reduction by FldHQ and monitored by the formation of FldSQ at 580 nm. Av2 addition occurs at 9 s. Two electron equivalents are transferred to [Fe₄S₄]²⁺ Av2 (top curve), one to [Fe₄S₄]¹⁺ Av2 (middle curve), and none to S = 4 [Fe₄S₄]⁰ (bottom curve). A control for electron equivalents, [Fe(CN)₆]^3–, also is shown. In the presence of 1 mM MgATP, reduction is 79% efficient in the trace shown, although the average of several runs was ≈73% efficient. Varying degrees of slow oxidation of FldHQ are observed, most notably in the curve for S = 4 [Fe₄S₄]⁰ addition.

The first possibility has been suggested in ref. 18. If the E_m of 1+/0 is –790 mV, then an Av2 conformational change with FldHQ would have to increase its E_m to form [Fe₄S₄]⁰ Av2. Alternatively, if the E_m of 1+/0 is –460 mV, then an Av2 conformational change in the presence of certain nonphysiological reductants would tend to decrease its E_m, making it more difficult for these reductants to reduce Av2. There is no precedent for an Av2 conformational change increasing its E_m. Rather, Av2 conformational changes, upon binding of nucleotides and Av1, for instance, tend to decrease the E_m of Av2 (1, 18). So, the true E_m of 1+/0 is closer to –460 mV, making it possible for FldHQ to reduce Av2 to [Fe₄S₄]⁰ Av2, yet making it more difficult for nonphysiological reductants to form [Fe₄S₄]⁰ Av2.

The second general possibility is that the reduction potential for Fld may change by conformational change in a complex with Av2. It was shown that in Klebsiella pneumoniae, a complex forms between KpFldSQ and [Fe₄S₄]¹⁺ Kp2(MgAXP)₂ but not with nucleotide-free [Fe₄S₄]¹⁺ Kp2 (24). Similar results are seen in Rhodobacter capsulatus (29). If these proteins in A. vinelandii interact in homologous fashion to those in K. pneumoniae and R. capsulatus, then it is not likely that complex formation occurs between nucleotide-free Av2 and AvFld. Therefore, a change in the reduction potential for Fld in a complex with Av2 is not likely to explain the data in Fig. 1.

Whereas the rapid kinetics in Fig. 1 suggests the formation of [Fe₄S₄]⁰ Av2, there is also a slow formation of FldSQ in some traces. FldHQ is stable in the absence of O₂, so slow formation of FldSQ only occurs upon Av2 addition. An extreme example is shown in the lower curve of Fig. 1, where Ti(III)-reduced [Fe₄S₄]⁰ Av2 is added to FldHQ. When buffer is added in place of Av2, little FldSQ is formed, ruling out nonspecific oxidation. Gas chromatography revealed that no H₂ was formed. Pretreating all solutions with chelex resin diminished the slow rate of FldHQ oxidation, suggesting that free metal ions from inactive or partially damaged Av2 may be partially responsible for slow FldHQ formation.

Optical Spectrum of FldHQ-Reduced [Fe₄S₄]⁰ Av2.

[Fe₄S₄]¹⁺ Av2 (63 kDa) was reacted with excess FldHQ (21 kDa), and the resulting mixture was separated anaerobically by using a G-75 Sephadex column (1.0 × 50 cm). The spectrum of FldHQ-reduced [Fe₄S₄]⁰ Av2 resembled the spectrum of [Fe₄S₄]¹⁺ Av2, which does not have the shoulder at 520 nm typical of Ti(III)-reduced [Fe₄S₄]⁰ Av2. Because these samples were dilute and prone to error in spectral measurements, the optical spectrum of FldHQ-reduced [Fe₄S₄]⁰ Av2 also was obtained through difference spectroscopy, which required two experiments. First, [Fe₄S₄]¹⁺ Av2 was added to excess FldHQ, forming [Fe₄S₄]⁰ Av2, and the spectrum recorded vs. the original spectrum of FldHQ, correcting for dilution. In a similar fashion, a control spectrum was recorded by adding an identical amount of [Fe(CN)₆]^3– to an identical FldHQ sample, forming [Fe(CN)₆]^4–. The difference between these two spectra represents the spectrum of FldHQ-reduced [Fe₄S₄]⁰ Av2 minus an equimolar amount of [Fe(CN)₆]^4– that has negligible absorbance in the visible range. The resulting spectrum in Fig. 2 is consistent with that obtained from the dilute Fld-free [Fe₄S₄]⁰ Av2 separated by Sephadex G-75 chromatography discussed above and by MV reduction (4). The spectrum resembles that of [Fe₄S₄]¹⁺ Av2 but, importantly, lacks the shoulder at 520 nm present in S = 4 [Fe₄S₄]⁰ Av2 (see Fig. 2), suggesting that [Fe₄S₄]⁰ Av2 formed by FldHQ is different from that formed by Ti(III).

Fig. 2.

Determination of the spectrum of Av2 in the S = 0 [Fe₄S₄]⁰ state. Two curves show the difference in absorbance in a sample of FldHQ before and after addition of either [Fe(CN)₆]^3– or [Fe₄S₄]¹⁺ Av2. The difference between these curves yields the absorbance spectrum of 43 μM Av2 in the S = 0 [Fe₄S₄]⁰ state (minus the near-zero absorbance of an equivalent concentration of [Fe(CN)₆]^4– indicated by the arrow). For comparison, the absorbance spectrum of Ti(III)-reduced [Fe₄S₄]⁰ Av2 is shown at an equivalent concentration with its characteristic shoulder at 520 nm.

EPR-Silent [Fe₄S₄]⁰ Av2.

Parallel mode EPR spectra of Ti(III)- and FldHQ-reduced [Fe₄S₄]⁰ Av2 at the same concentration are compared in Fig. 3. The Ti(III)-reduced [Fe₄S₄]⁰ Av2 has a g = 16.4 signal characteristic of the S = 4 spin state. FldHQ-reduced [Fe₄S₄]⁰ Av2 lacks this g = 16.4 signal.

Fig. 3.

Parallel-mode EPR spectra of [Fe₄S₄]⁰ Av2 proteins. The signal from 0.05 mM Ti(III)-reduced S = 4 [Fe₄S₄]⁰ Av2 with its characteristic g = 16.4 signal is contrasted with an equivalent concentration of FldHQ-reduced S = 0 [Fe₄S₄]⁰ Av2.

The perpendicular-mode EPR spectra of equimolar [Fe₄S₄]¹⁺ Av2 and [Fe₄S₄]⁰ Av2 plus a 2-fold excess of FldHQ are compared in Fig. 4. The sample containing both Fld and Av2 shows some overlap between the g = 1.94 signal from [Fe₄S₄]¹⁺ Av2 (30) and the very intense g = 2.00 signal of FldSQ (28). Still, the intensity of the [Fe₄S₄]¹⁺ Av2 g = 1.94 signal decreased 70–80% during reduction by FldHQ to [Fe₄S₄]⁰ Av2. A similar result was observed in ref. 4 by using MV (E_m = –460 mV) to reduce [Fe₄S₄]¹⁺ Av2 to [Fe₄S₄]⁰ Av2. This EPR result and the parallel mode EPR result are consistent with the FldHQ-reduced [Fe₄S₄]⁰ Av2 being in an EPR-silent S = 0 spin state.

Fig. 4.

Perpendicular-mode EPR spectra for the reduction of [Fe₄S₄]¹⁺ Av2 by FldHQ. A sample containing 0.16 mM [Fe₄S₄]¹⁺ Av2 plus 0.33 mM FldHQ (blue) is consistent with 70–80% reduction of [Fe₄S₄]¹⁺ Av2 to S = 0 [Fe₄S₄]⁰ Av2 when compared with both a 0.16 mM [Fe₄S₄]¹⁺ Av2 sample (brown) and a 0.16 mM FldSQ sample prepared by the addition of [Fe(CN)₆]^3– to FldHQ (data not shown). A more effective means of preparing S = 0 [Fe₄S₄]⁰ Av2 (89% fully reduced) involved loading a mixture of DT, Av2, and FldHQ together on a size-exclusion column for separation from DT and collection of the protein mixture effluent (aqua). In this trace, the signal of FldSQ is circled.

A more effective method to prepare S = 0 [Fe₄S₄]⁰ Av2 is first to combine DT, Av2, and Fld, allow them to react for ≈10 min, and then separate the protein mixture by anaerobic Sephadex G-25 gel filtration. A representative EPR spectrum is shown in Fig. 4 and is consistent with 89% reduction of [Fe₄S₄]¹⁺ to [Fe₄S₄]⁰ Av2. Little FldSQ is present in this sample, as demonstrated by the small g = 2 signal due to FldSQ.

Evans NMR Magnetic Susceptibility.

The Evans NMR method was used to determine the spin state of Av2 in various redox states. A direct correlation exists between the changes in chemical shift of a reference compound with the electronic spin state of a substance. Fig. 5 shows how this technique is applied to various oxidation states of Av2 and FldSQ. The upper line, with a slope of 37.1 Hz/mM, is for Ti(III)-reduced S = 4 [Fe₄S₄]⁰ Av2. With a slope of 11.2 Hz/mM, [Fe₄S₄]¹⁺ Av2 consists of a mixture of S = 1/2 and S = 3/2. FldSQ has a spin of S = 1/2, with a slope of 4.6 Hz/mM; and [Fe₄S₄]²⁺ Av2 has a spin of S = 0 with a slope of 0 Hz/mM. The ratio of the nonzero slopes are 8.1/2.4/1.0, which correlates with the number of unpaired electrons of each species. Ti(III)-reduced [Fe₄S₄]⁰ Av2 (S = 4) has a spin 8.1 times that of FldSQ (S = 1/2). [Fe₄S₄]¹⁺ Av2 has an overall spin 2.4 times that of FldSQ, consistent with a mixture of 47% S = 1/2 and 53% S = 3/2. These results show that the Evans NMR method gives reliable results for paramagnetic proteins in known magnetic states.

Fig. 5.

Magnetic susceptibility of Fe protein and FldSQ. Determined by the proton shift in reference solution as a function of protein concentration, according to the Evans NMR method. The slopes for different protein species are proportional to their spin state. Ti(III)-reduced [Fe₄S₄]⁰ Av2 (S = 4) has a slope 8.1 times larger than that of FldSQ (S = [1/2]) and 2.4 times that of [Fe₄S₄]¹⁺ Av2 (S = [1/2] and S = 3/2). [Fe₄S₄]²⁺ Av2 has a zero slope (S = 0). The result for FldHQ-reduced [Fe₄S₄]⁰ Av2 is consistent with [Fe₄S₄]⁰ being in the S = 0 spin state.

Addition of [Fe₄S₄]¹⁺ Av2 to FldHQ produces combined paramagnetic shifts arising from the formation of FldSQ and [Fe₄S₄]⁰ Av2. The contribution of [Fe₄S₄]⁰ Av2 was determined by subtracting the contribution of FldSQ, given its concentration determined optically times its slope (hertz/micromole) in the standard curve for proton shift in Fig. 5. The residual shift is due to the magnetic contribution of FldHQ-reduced [Fe₄S₄]⁰ Av2. This result is shown as a single point in Fig. 5, an average of 0.8 ± 1.4 Hz (1.0, −0.50, 0.0, and 2.6 Hz) at 0.3 mM, which corresponds to a slope of 2.7 Hz/mM. The deviation among these data arises from errors in measuring FldSQ concentration determined optically before NMR measurements, allowing time for slow oxidation of FldHQ to make these NMR shifts more positive than they should be. Incomplete reduction of [Fe₄S₄]¹⁺ Av2 to [Fe₄S₄]⁰ Av2 also could contribute to larger shifts. Neglecting slow FldSQ formation, this average slope of 2.7 Hz/mM rules out formation of S = 4 [Fe₄S₄]⁰ Av2 and is consistent with a lower limit of 76% reduction to S = 0 [Fe₄S₄]⁰ Av2. Oxidation of [Fe₄S₄]¹⁺ Av2 to [Fe₄S₄]²⁺ Av2 (S = 0) during the NMR measurements would contribute to a lower NMR shift, accounting for this Evans NMR result for FldHQ plus Av2. However, this possibility is unlikely because the Evans NMR measurements were done under strongly reducing conditions. Rather, the Evans NMR data are consistent with the formation of S = 0 [Fe₄S₄]⁰ Av2.

Effect of Nucleotides.

The physiological relevance of [Fe₄S₄]⁰ Av2 is supported by the characterization of the previously undescribed form of [Fe₄S₄]⁰ Av2 made by reduction with the physiological reductant FldHQ. However, a complete description of nitrogenase catalysis by using [Fe₄S₄]⁰ Av2 requires evaluation of the reduction of nucleotide-bound forms of Av2, because these are catalytically relevant (1, 31). Nucleotides decrease the reduction potential for the 2+/1+ couple (1, 32–36), although it is not yet known to what degree this behavior also occurs for the 1+/0 couple. The L127Δ Av2 mutant locked into an MgATP-bound conformation (33) can be reduced to the all-ferrous state by Ti(III) citrate (W. N. Lanzilotta and L. C. Seefeldt, personal communication). This result is interesting considering that the E_m for the L127Δ 2+/1+ redox couple is the same as Av2 wild-type bound to MgATP (33). Nucleotide-bound Av2 also must be reduced to [Fe₄S₄]⁰ Av2(MgAXP)₂ by Ti(III) because: (i) Ti(III) operates the 2+/0 redox couple with ATP/2e^– = 2; and (ii) reduction of nucleotide-bound Av2 is catalytically relevant (see Scheme 1). This reasoning also holds when FldHQ is the reductant, so presumably FldHQ makes [Fe₄S₄]⁰ Av2(MgAXP)₂. This behavior is not surprising because E_m = –515 mV for FldHQ is comparable to E_m = –510 mV reported for Ti(III).

Scheme 1.

The Fe protein cycle. Incorporated are conventional catalysis pathways (solid arrows) and pathways unique to the catalysis of the all-ferrous Fe protein (dotted arrows), where E_n is half of MoFe protein (one active site) reduced by n electrons beyond the DT-reduced state. Note the absence of reduction of nucleotide-free [Fe₄S₄]²⁺ to [Fe₄S₄]¹⁺ and [Fe₄S₄]⁰, which are not significant pathways during catalysis because of the strong binding of nucleotides.

Av2 reduction by FldHQ was repeated in the presence of 1.0 mM MgADP and MgATP (see Fig. 1). Rapid electron transfer to Av2 readily occurred. However, whereas reduction of nucleotide-free [Fe₄S₄]²⁺ Av2 and [Fe₄S₄]¹⁺ Av2 was a near-integer value, only ≈73% of [Fe₄S₄]²⁺ or [Fe₄S₄]¹⁺ Av2(MgATP)₂ was reduced to [Fe₄S₄]⁰ Av2(MgATP)₂ rapidly. The presence of ADP was even more inhibitory because only ≈40% of [Fe₄S₄]²⁺ or [Fe₄S₄]¹⁺ Av2(MgADP)₂ was rapidly reduced to [Fe₄S₄]⁰ Av2(MgADP)₂.

All-ferrous activity (ATP/2e^– = 2) is observed with FldHQ as the reductant. If incomplete reduction were to occur, then we should observe more character of the 2+/1+ couple (ATP/2e^– ≥ 4). For example, if only 73% of the total pool of Av2 were reduced to [Fe₄S₄]⁰ Av2(MgAXP), then that would yield an ATP/2e^– ratio of 2.54, measurably higher than the observed lower limit of 2.0, suggesting that the remaining Av2(MgAXP)₂ may be reduced to the [Fe₄S₄]⁰ state in slower phases of reduction. In particular, nucleotides are known to lower the rate of reduction of Av2 by DT (37–42), even breaking up the reduction into multiple phases, some fast, some slow.

Still, it is possible that the FldSQ/FldHQ couple may not be sufficient for stoichiometric conversion to [Fe₄S₄]⁰ Av2(MgAXP)₂, given a drop in reduction potential of Av2 upon binding nucleotides. However, the FldHQ/Av2 ratio was ≈10 in the presence of nucleotides, and if the incomplete reduction of [Fe₄S₄]¹⁺ Av2(MgAXP)₂ by FldHQ is due to a shift in potential, this shift is quite significant. The inhibitory effect of nucleotides on the extent of rapid reduction may be a combination of kinetic and thermodynamic effects, and measurements of the potential shift of the E_m for the 1+/0 couple in the presence of nucleotides should evaluate these possibilities.

Physiological Relevance of [Fe₄S₄]⁰ Av2.

The evidence that [Fe₄S₄]⁰ Av2 is formed by its in vivo reductant has strong implications for the physiological relevance of [Fe₄S₄]⁰ Av2. Even so, it may be that nitrogenase catalysis proceeds with both 2+/0 and 2+/1+ redox couples simultaneously, or perhaps one or the other predominates depending on in vivo conditions, such as temperature, pH, ionic strength, FldHQ concentration, ATP/ADP, and component protein concentrations. Some of these considerations are summarized in Scheme 1, an adapted Fe protein cycle (43) to account for all-ferrous activity.

There are two likely pathways for the formation of [Fe₄S₄]⁰ Av2(MgAXP)₂ in vivo, either by reduction of [Fe₄S₄]¹⁺ Av2(MgADP)₂ or reduction of [Fe₄S₄]¹⁺ Av2(MgATP)₂. Nucleotide exchange is fairly rapid after reduction of [Fe₄S₄]²⁺ to [Fe₄S₄]¹⁺ (43). Therefore, reduction of [Fe₄S₄]¹⁺ Av2(MgADP)₂ to the [Fe₄S₄]⁰ state would be favored under conditions of limiting ATP (high ADP/ATP). However, limiting ATP in vivo coincides with a state of energy starvation with a low FldHQ/FldSQ ratio, which would make it more difficult to form the [Fe₄S₄]⁰ state of Av2. Interestingly, this study has shown that only 40% of [Fe₄S₄]¹⁺ Av2(MgADP)₂ is rapidly reduced to [Fe₄S₄]⁰ Av2(MgADP)₂ by an excess of FldHQ.

On the other hand, reduction of [Fe₄S₄]¹⁺ Av2(MgATP)₂ would be favored under conditions of excess ATP (high ATP/ADP), which would coincide with active electron transport (high FldHQ/FldSQ) to nitrogenase. However, binding of Av1 to [Fe₄S₄]¹⁺ Av2 (MgATP)₂ would compete with [Fe₄S₄]¹⁺ Av2(MgATP)₂ reduction by FldHQ, so reduction would have to be relatively efficient to compete effectively with the 2+/1+ redox couple. Coincidentally, this study has shown that ≈70–80% of [Fe₄S₄]¹⁺ Av2(MgATP)₂ is rapidly reduced to [Fe₄S₄]⁰ Av2(MgATP)₂ by FldHQ, or nearly twice as much [Fe₄S₄]⁰ production in the presence of ATP as opposed to ADP. Thus, in vivo production of [Fe₄S₄]⁰ Av2 would most likely occur under the conditions already known to favor efficient nitrogenase catalysis: a high Av2/Av1 ratio, a high FldHQ/Av2 ratio, and a high ATP/ADP ratio.

Summary and Conclusion.

Both Ti(III) citrate and FldHQ support efficient nitrogenase catalysis yielding ATP/2e^– = 2. The Ti(III)-reduced S = 4 [Fe₄S₄]⁰ Av2 has unique properties that distinguish it from [Fe₄S₄]²⁺ and [Fe₄S₄]¹⁺ Av2. However, these properties were not detected by using the in vivo reductant FldHQ, and objections were raised as to whether FldHQ forms [Fe₄S₄]⁰ Av2. However, such objections did not take into account that FldHQ-reduced [Fe₄S₄]⁰ Av2 is in a different spin state with properties that make it difficult to distinguish from [Fe₄S₄]²⁺ and [Fe₄S₄]¹⁺ Av2. Specifically, the combination of EPR, UV-vis, and Evans NMR results are consistent with the formation of a previously uncharacterized [Fe₄S₄]⁰ Av2 that is in an S = 0 spin state like [Fe₄S₄]²⁺ Av2, yet has an optical spectrum that resembles that of [Fe₄S₄]¹⁺ Av2.

This evidence renews the debate over the physiological relevance of the all-ferrous Fe protein and calls to question the more recently measured E_m = –790 mV for Cr(II)-EDTA-reduced [Fe₄S₄]⁰ Av2. That both MV and FldHQ appear capable of producing the [Fe₄S₄]⁰ Av2 suggests that some reductants, i.e., Cr(II)-EDTA, may artificially lower the measured E_m for [Fe₄S₄]⁰ Av2 formation, possibly through a reductant-induced protein conformational change.

Even so, the physiologically relevant forms of Av2 are nucleotide-bound species, which are less readily reduced by FldHQ than is nucleotide-free Av2. But MgATP appears to be only slightly inhibitory (70–80% rapid reduction of [Fe₄S₄]¹⁺ Av2(MgATP)₂ by FldHQ), suggesting that all-ferrous ATP/2e^– = 2 activity could be possible in vivo under high-activity conditions, i.e., excess Av2/Av1, high FldHQ/Av2, and high ATP/ADP. This conclusion is supported by the observation that the ATP/2e^– ratio falls to 2.0 during catalysis with FldHQ in vitro (5).

Methods

Anaerobic Procedures.

Reactions involving air-sensitive compounds were conducted in an argon-filled Vacuum Atmospheres (Hawthorne, CA) glove box equipped with a Nyad O₂ monitor and dual purifiers (O₂ < 0.1 ppm). Optical spectra were recorded on an HP 8453 diode array spectrometer located inside the glove box.

Reagent Preparation.

Av2 was isolated and characterized as reported in ref. 44 with activities of 1,885–2,030 nmol H₂·min^–1·mg of Av2^–1. Ti(III) citrate was prepared according to established protocols in refs. 7 and 17. AvFld was isolated during the nitrogenase isolation and purified by using an ultragel size-exclusion column (5 cm × 2.0 m). FldHQ was prepared by reduction of Fld with excess DT in 0.05 M Tris/0.1 M NaCl, pH 8.0 and separated from DT on a Sephadex G-50 column (1.0 × 20 cm) equilibrated with 0.05 M Tris, pH 8.0. Treating all solutions with a chelex (Sigma, St. Louis, MO) chelating column and treating all glassware and syringes with sulfuric acid prolonged the lifetime of FldHQ.

Reduction of Av2 by FldHQ.

FldHQ from 0.25 to 0.50 mM was added to a 1.0-ml quartz cell, and the stable, near-zero absorbance at 580 nm was monitored for 1–5 min. Solutions of DT-free [Fe₄S₄]²⁺ Av2, [Fe₄S₄]¹⁺ Av2, or [Fe₄S₄]⁰ Av2 (at a 3- to 20-fold excess of FldHQ) were added by syringe, and the solution was mixed. Reaction was monitored optically for the conversion of FldHQ to FldSQ (Δλ_max = 580 nm, Δε = 5.4 mM^–1·cm^–1; refs. 21 and 28) and compared with controls with only the addition of anaerobic buffer. From the optical change and the known amounts of Fld and Av2, the stoichiometry of electron transfer was determined. Identical reactions were conducted in the presence of FldHQ containing 1.0 mM MgATP and MgADP.

EPR Analysis.

EPR samples of DT-, FldHQ-, and Ti(III)-reduced Av2 were prepared in the glove box in 3.0-mm internal diameter-calibrated quartz EPR tubes, capped, and transferred outside where they were frozen in liquid nitrogen. EPR spectra were recorded in both perpendicular and parallel modes at 12–15 K with 10–50 scans per spectrum by using a 9.2 GHz Bruker EMX EPR spectrometer with a dual cavity.

S = 4 [Fe₄S₄]⁰ Av2 was prepared by reduction with Ti(III) citrate, and excess Ti(III) was removed by anaerobic G-25 Sephadex chromatography. EPR spectra of [Fe₄S₄]⁰ Av2 at g = 16.4 over the concentration range of 0.05–0.25 mM were examined to determine the detection limit of ≈5.0 μM for Ti(III)-reduced [Fe₄S₄]⁰ Av2.

EPR samples of FldHQ-reduced [Fe₄S₄]⁰ Av2 were prepared from [Fe₄S₄]¹⁺ Av2 (0.15–0.4 mM) plus an excess of FldHQ (2- to 10-fold excess). Independent EPR samples forming FldSQ were prepared by the addition of [Fe(CN)₆]^3– to FldHQ. Alternatively, EPR samples were prepared by first mixing DT, Av2, and Fld, then separating the protein mixture from DT on a G-50 Sephadex column (1.0 × 20 cm) equilibrated with 0.05 M Tris, pH 8.0. The DT-free protein mixture was used for EPR measurements.

Evans NMR Magnetic Susceptibility.

Magnetic susceptibility measurements for standard metal ion or protein solutions (0.25–10 mM) of known spin state were determined by the Evans NMR method (45) with an INOVA 300 MHz or a Varian 500 MHz spectrometer. The reference solution in the capillary insert and the sample solution in the 5 mm NMR tube was 20% ²H₂O/40 mM N-tris[hydroxymethyl]methyl-2-aminoethanesulfonic acid, pH 7.4. Tertiary butyl alcohol, t-BuOH, was added to the reference and sample solutions for a final composition of 7% and 0.6%, respectively. Additional experiments without t-BuOH used Tris both as buffer and internal standard for proton resonance. The chemical shift of each protein sample was calculated by subtracting the measured proton shift from the corresponding capillary reference. Protein standard curves were constructed by using [Fe₄S₄]²⁺ Av2 (S = 0), [Fe₄S₄]¹⁺ Av2 (a mixture of S = [1/2] and S = 3/2), FldSQ (S = [1/2]), and Ti(III)-reduced [Fe₄S₄]⁰ Av2 (S = 4). The slope in Hz/mM obtained by using Tris or t-BuOH as frequency references for the protein or metal ion standards were identical.

Additional NMR samples were prepared under identical conditions as above except [Fe₄S₄]¹⁺ Av2 was added to a 3–5 molar excess of FldHQ to give a final concentration of ≈0.30 mM Av2. Independent NMR controls were prepared by the addition of [Fe(CN)₆]^3– to FldHQ for comparison.

Abbreviations

Av1

MoFe protein from Azotobacter vinelandii

Av2

Fe protein from Azotobacter vinelandii

Fld

flavodoxin in an arbitrary oxidation state

FldSQ

radical singly reduced flavodoxin semiquinone

FldHQ

flavodoxin hydroquinone

MV

methyl viologen

DT

sodium dithionite