Nitrogenase is a globally important enzyme that catalyses the reduction of atmospheric dinitrogen into ammonia and is thus an important part of the nitrogen cycle. The nitrogenase enzyme is composed of a catalytic molybdenum–iron protein (MoFe protein) and a protein containing an [Fe4–S4] cluster (Fe protein) that functions as a dedicated ATP-dependent reductase. The current understanding of electron transfer between these two proteins is based on stopped-flow spectrophotometry, which has allowed the rates of complex formation and electron transfer to be accurately determined. Surprisingly, a total of four Fe protein molecules are required to saturate one MoFe protein molecule, despite there being only two well-characterized Fe-protein-binding sites. This has led to the conclusion that the purified Fe protein is only half-active with respect to electron transfer to the MoFe protein. Studies on the electron transfer between both proteins using rapid-quench EPR confirmed that, during pre-steady-state electron transfer, the Fe protein only becomes half-oxidized. However, stopped-flow spectrophotometry on MoFe protein that had only one active site occupied was saturated by approximately three Fe protein equivalents. These results imply that the Fe protein has a second interaction during the initial stages of mixing that is not involved in electron transfer.
The nitrogenase catalytic cycle
Nitrogen-fixing bacteria are responsible for the reduction of over 100 million tonnes of atmospheric dinitrogen globally . The bacteria that fix dinitrogen are ubiquitous, ranging from free living aerobic (e.g. Azotobacter vinelandii) and anaerobic bacteria (e.g. Clostridium pasteurianum) to symbiotes living in root nodules of plants (e.g. Bradyrhizobium japonicum) . Although the organisms that express nitrogenase are diverse, the functional enzyme is genetically well conserved. Three types of nitrogenase have been properly characterized: a molybdenum–iron enzyme, a vanadium–iron enzyme and an iron-only enzyme. Nitrogen-fixing bacteria often have the genes for more than one nitrogenase and expression is carefully regulated . The Mo-containing nitrogenase is preferentially expressed; however, in the absence of Mo the V nitrogenase is expressed and, in the absence of both, the Fe-only form is expressed. Of the different types of nitrogenase, the Mo nitrogenase is the most active and has been at the centre of intensive research over the last three decades . The Mo nitrogenase reduces dinitrogen to ammonia via the following reaction: N2+8H++8e−+16MgATP→2NH3+H2+16MgADP+16Pi
The Mo nitrogenase comprises two component proteins: an Fe protein (reductase) responsible for MgATP-driven electron transfer, and an MoFe protein that receives electrons and contains the active site for dinitrogen reduction. X-ray crystallography studies have revealed the structure of the MoFe protein, Fe protein and complexes of the two proteins together [4–6]. The MoFe protein is an α2β2 heterotetrameric protein that contains two unique cofactors, an Fe8S7 complex known as the P-cluster, and the active-site cofactor, an iron–molybdenum–homocitrate complex known as FeMoco (FeMo cofactor) . The Fe protein is a homodimeric protein containing an [Fe4–S4] cluster and two MgATP-binding sites . The two halves of the Fe protein are flexible and adopt different conformations during ATP binding, complex formation and ATP hydrolysis . The conformational change that occurs after association of the Fe protein with MgATP affects both the redox potential of the [Fe4–S4] cluster and increases the affinity of the Fe protein for the MoFe protein. The MgATP-associated Fe protein forms a complex with the MoFe protein that further lowers the redox potential of the Fe protein [Fe4–S4] cluster. It is at this point that electron transfer from the Fe protein to the MoFe protein is proposed to occur (Figure 1). Hydrolysis of ATP also occurs and results in complex dissociation [8,9]. It is clear that ATP-dependent electron transfer is central to the nitrogenase mechanism as the MgATP-bound Fe protein is the only known reductant capable of reducing the MoFe protein to a state that supports dinitrogen reduction .
The Lowe–Thorneley Fe protein cycle of K. pneumoniae
Pre-steady-state analysis of electron transfer in nitrogenase
Anaerobic stopped-flow spectrophotometry has been used to study the electron transfer event between the Fe protein and MoFe protein. Mixing the reduced Fe protein with the MoFe protein and MgATP causes a rapid oxidation of the Fe protein that can be observed as an increase in absorbance at 430 nm, whereas the corresponding reduction of the MoFe protein does not cause an observable spectrophotometric change. The rate of Fe protein oxidation is well characterized and gives a pseudo-first-order rate constant of approx. 150–200 s−1 when the Fe protein is saturated with MoFe protein [11,12]. This value is similar to the first-order rate of electron transfer and coupled ATP hydrolysis within the Fe–MoFe protein complex (Figure 1). Surprisingly, although purified samples of the Fe protein can be fully cycled between oxidized and reduced forms using chemical reductants and oxidants, stopped-flow spectrophotometric titration of the MoFe protein with increasing amounts of Fe protein in the presence of excess MgATP reach saturation at an approximate MoFe protein/Fe protein ratio of 4. This has been attributed to the Fe protein being approx. 50% active in supporting MgATP-dependent electron transfer to the MoFe protein .
Another technique used to study electron transfer is rapid-freeze EPR. Both the MoFe protein and Fe protein have distinctive paramagnetic spectra in the dithionite-reduced state. At pH 7.4 and in the presence of the reductant dithionite, the MoFe protein from Klebsiella pneumoniae has an EPR spectrum of a rhombic S=3/2 system with g values of 4.32, 3.65 and 2.00 optimally observed under EPR conditions of 10 K and 100 mW . The Fe protein has an S=1/2 signal with g values of 2.05, 1.94 and 1.87 optimally observable at 18 K and 10 mW; upon binding of MgATP, the signal changes from a rhombic form to an axial form with g values of 2.04 and 1.93 . The signal features of the two proteins are easily identifiable in the EPR spectra and it is therefore possible to study the oxidation state of both proteins during turnover.
Rapid-freeze EPR was first applied to nitrogenase as a method of studying the initial steps of dinitrogen reduction at Fe protein/MoFe protein molar ratios of 2:1 in the presence of an ATP-regeneration system. However, the results were difficult to reconcile. At the time, the difficulties involved in preparing samples for EPR and interpreting the observed changes in signal coupled with the requirement for large amounts of pure nitrogenase component proteins prohibited the routine use of rapid-freeze EPR as a tool for nitrogenase research. Very little rapid-freeze EPR work has been reported on nitrogenase since these early experiments, so in order to investigate the interaction between the Fe protein and MoFe protein by rapid-freeze EPR we purified gram quantities of highly active nitrogenase component proteins from K. pneumonia . The specific activity of the pure Fe protein was 1900 nmol of C2H2 reduced/min per mg, whereas the specific activity of the MoFe protein was 2500 nmol of C2H2 reduced/min per mg and contained 2.0 Mo·(MoFe protein)−1. This indicated that both halves of the MoFe protein contained the FeMoco. Rapid-freeze experiments were carried out as described previously . Both syringes contained a buffer of 50 mM Hepes (pH 7.4), 10 mM MgCl2 and 10 mM Na2S2O4. Syringe 1 contained the MoFe protein and Fe protein at different molar ratios, whereas syringe 2 contained a stock solution of ATP. Both syringes were prepared in an anaerobic chamber and connected to the rapid-freeze apparatus under a stream of oxygen-free nitrogen. The oxygen scavenger and reductant Na2S2O4 was present at 10 mM in all buffers to ensure the nitrogenase component proteins were maintained under anaerobic conditions. After mixing for different time periods, the samples were quenched in 2-methylbutane at 130 K, packed into EPR tubes and the EPR spectra was measured for each sample at 18 K and 10 mW.
Figure 2 shows the EPR spectra of rapid-freeze sample experiments, where the final concentration of mixed protein was 9 mM MgATP, 322 μM MoFe protein and 164 μM Fe protein, giving a molar ratio of 1 MoFe protein and 0.5 Fe protein. To obtain a 0 ms time point, the proteins were mixed with buffer without ATP, whereas time intervals between 5 and 108 ms were obtained by mixing component proteins with the stock ATP. During the first 5 ms, the EPR signal corresponding to the [Fe4–S4] cluster of the Fe protein showed a rapid decrease of approx. 50%, after which it remained relatively constant. A subtle change in the spectrum reflecting the conformation of MgATP-bound Fe protein was also observed. Both signals were quantified by double integration and expressed as a percentage of initial signal intensity (Figure 2B). The observed oxidation of Fe protein suggests that the majority of electron transfer from Fe protein to MoFe protein occurs within 5 ms, although this is much faster than would be predicted from the 200 s−1 electron transfer rate from stopped-flow spectrophotometry. The initial loss of 10% signal intensity in the MoFe protein in the first 5 ms is directly related to the amount of Fe protein oxidized, and the intensity of the FeMoco signal continues to decrease as successive cycles of reduction by the Fe protein occur.
Rapid-freeze experiments on nitrogenase components measured using MoFe protein/Fe protein ratios of 1:4 or 1:0.5
Rapid freeze was also performed at a MoFe protein/Fe protein ratio of 1:4, where the saturation of the MoFe protein by the Fe protein using stopped-flow spectrophotometry is observed. After mixing, the MoFe protein concentration was 100 μM, the Fe protein concentration was 400 μM and the concentration of ATP was 9 mM. The final buffer composition was 25 mM Hepes (pH 7.4), 10 mM MgCl2 and 10 mM Na2S2O4. As observed at the previous ratio of 1:0.5, the initial rate of FeMoco reduction is directly comparable with the rate of oxidation of Fe protein. Surprisingly, these results suggest that the MoFe protein is not completely saturated under these conditions, as there is a residual MoFe protein signal of approx. 30% after 100 ms. The decrease in EPR spectra of [Fe4–S4] is 45%, consistent with the Fe protein being only half-active.
Stopped-flow spectrophotometry of MoFe protein with only one active site
It is possible to separate the MoFe protein of K. pneumonia into two species during purification by ion-exchange chromatography; one where both active sites are occupied by FeMoco and with an Mo content of 2.0 Mo·(MoFe protein)−1, and one where only one active site is occupied and containing 1.0 Mo·(MoFe protein)−1 . The specific activity of purified samples of the fully occupied MoFe protein was measured as 2300 nmol of C2H2 reduced/min, whereas pure samples of the half-occupied MoFe protein have a specific activity of 1200 nmol of C2H2 reduced/min. The specific activity of the Fe protein used was 1200 nmol of C2H2 reduced/min. Stopped-flow spectrophotometry was performed on both the fully occupied and half-occupied forms of MoFe protein to observe any differences that might occur in the initial electron transfer reaction. Mixing either of the two MoFe protein forms and a 7-fold excess of Fe protein with 10 mM MgATP cause a rapid increase in absorbance at 430 nm to be observed. The amplitude of the absorbance change was greater in the case of the fully occupied MoFe protein, but the rate of Fe protein oxidation was similar for both fully occupied and half-occupied MoFe protein with pseudo-first-order rate constants of 110 s−1 and 130 s−1 respectively (Figure 3A). These rates are comparable with the published value of 140 s−1 for electron transfer between the component proteins of K. pneumoniae . These data show that the rate of electron transfer through the MoFe protein–Fe protein complex is not significantly affected by the absence of metal cofactors on the opposing αβ-subunit dimer of the MoFe protein.
Stopped-flow spectrophotometry of Fe protein with MoFe protein with one or both FeMoco clusters incorporated
Titration of either form of MoFe protein with increasing concentrations of Fe protein in the presence of MgATP and Na2S2O4 causes a corresponding increase in amplitude at 430 nm. Initially, when the MoFe protein concentration was greater than the Fe protein concentration, the amplitude of the absorbance change attributed to the oxidation of the Fe protein [Fe4–S4] cluster was the same for both forms of the MoFe protein species. However, as the Fe protein concentration increased, the change in amplitude became constant as the MoFe protein became saturated. The MoFe protein with only one occupied active site became saturated at a Fe protein/MoFe protein molar ratio of approx. 3.5:1, whereas the fully occupied MoFe protein became saturated at the previously reported Fe protein/MoFe protein molar ratio of 4.5:1, consistent with a greater proportion of functional Fe-protein-binding sites on fully occupied MoFe . The difference between the molar ratios is less than would be expected if the Fe protein was half-active with respect to electron transfer between both MoFe protein forms.
Taken together, these rapid-freeze EPR and stopped-flow spectrophotometric experiments provide some interesting new details about the interaction between the MoFe protein and the Fe protein. It is evident from the rapid-quench data that only half the available Fe protein is involved in the initial electron transfer event, irrespective of an excess of available binding sites on the MoFe protein. This is in agreement with the proposed half-activity of the Fe protein. However, the Fe protein has been shown to be 100% capable of forming a stable complex with MoFe in the presence of BeF3− and MgADP when the concentration of MoFe protein active sites is in excess, raising the intriguing possibility that the observed half-activity of the Fe protein is related to the interactions that occur between the MoFe protein and Fe protein during the initial electron transfer event . The stopped-flow spectrophotometry on half-occupied MoFe protein is in agreement with this: under conditions where the numbers of Fe protein association sites are in excess, the change in absorbance is the same for both forms of MoFe protein. As the concentration of Fe protein increases, the half-occupied MoFe protein becomes saturated at an Fe protein/MoFe protein ratio that is lower than the fully occupied MoFe protein. However, this ratio of 3.5:1 is higher than the predicted 2:1 ratio expected from Figure 1, suggesting that the interaction between the two proteins is more complex than the current reduction cycle model would suggest.
Biological implications of ATP-dependent electron transfer
Electron transfer is an essential requirement of many biological systems. Whereas the majority of electron transfer reactions occur spontaneously, there is a small but diverse range of bacterial enzymes that require an energy input in the form of ATP to move electrons into an active site. Originally identified as a central function in nitrogenase, the ATP-dependent electron transfer process has now been identified in a broad range of enzymes, including the 2-hydroxyacyl-CoA dehydratase , light-independent protochlorophyllide reductase  and benzene dearomatase . The mechanism by which these enzymes couple ATP hydrolysis and electron transfer are comparable. The catalytic enzyme containing the active site receives electrons from the reductase during an event that involves complexation of the two proteins and hydrolysis of MgATP. The reductase in each of the three systems is similar, being a dimeric protein with two MgATP-binding sites and a [Fe4–S4] cubane cofactor bridging the two subunits. The close structural parallels between the Fe protein of nitrogenase and the other ATP-dependent reductases make it a useful model system for elucidating the mechanism and energetics of ATP-dependent electron transfer.
The stopped-flow and rapid-quench data demonstrate that, when the two nitrogenase component proteins are mixed in pre-steady-state conditions, the interaction that occurs during complex formation and electron transfer is complex and cannot be fully explained by an assumption that the Fe protein is half-active. These results suggest that the role of the Fe protein is more complex than a simple electron shuttle. Stopped-flow spectrophotometry on mutant Fe protein strains from A. vinelandii showed differences in the way that electrons could be transferred from the Fe protein to the MoFe protein [21,22]. Titration of MoFe protein with increasing amounts of modified Fe protein caused a linear increase in amplitude as observed for the wild-type. However, the amount of Fe protein required to saturate the MoFe protein occurred at a molar ratio of 2:1, half that observed for the wild-type Fe protein, but suggesting that these proteins were 100% active with respect to electron transfer. These results imply a second function for the Fe protein during the first electron transfer that can be removed through genetic modification. It has been demonstrated previously that the [Fe4–S4] cluster of both the nitrogenase Fe protein and the 2hydroxyacyl-CoA dehydratase Fe protein have the ability to cycle between three redox states, allowing it to transfer up to two electrons at a time . The nitrogenase Fe protein has also been shown to have a role in the biosynthesis of metalloclusters in the functional MoFe protein . Thus it is possible that there is a poorly understood second interaction during the initial stages of electron transfer between the two component proteins of nitrogenase. Further work, including the use of recent technological advances in observing electron transfer on sub-millisecond timescales, would help to resolve the more subtle interactions that occur between the Fe protein and MoFe protein.
Enzymology and Ecology of the Nitrogen Cycle: A Biochemical Society Focused Meeting held at University of Birmingham, U.K., 15–17 September 2010. Organized and Edited by Jeff Cole (University of Birmingham, U.K.), Rosa María Martínez-Espinosa (University of Alicante, Spain), David Richardson (University of East Anglia, Norwich, U.K.) and Nick Watmough (University of East Anglia, Norwich, U.K.).
T.A.C. is supported by a Research Councils UK Fellowship and a John and Pamela Salter award.