Kinetics of nitrogenase from Azotobacter vinelandii

Nitrogenase has been the subject of many investigations since the early 1960's. The catalytic mechanism of nitrogenase is unique because it couples the transfer of electrons with the hydrolysis of MgATP. The details of the mechanism are still to be revealed. The work described in this thesis aimed at a better understanding of nitrogenase catalysis. Questions we asked ourselves were:

What is the role of MgATP in nitrogenase catalysis?, and:

What is the sequence of events when electrons are transferred from the Fe protein to the MoFe protein? We tried to answer these questions mainly by using rapid kinetic methods, studying presteady-state reactions in nitrogenase catalysis. A brief introduction to nitrogenase is given in Chapter 1. obs

The MgATP-dependent pre-steady-state proton production by nitrogenase was studied by monitoring the absorbance changes of a pH indicator, cresol red (Chapter 2). The release of protons ( k_obs ≈ 14 s ^-1) was observed after a delay of -50 ms after mixing of the nitrogenase proteins with MgATP. The MgATP-dependent electron transfer from the Fe protein to the MoFe protein ( k_obs ≈ 100 s ^-1) started immediately after mixing. No proton production with a rate comparable to or higher than the rate of electron transfer was observed. These observations correspond to those of Mensink et al. (1992). The extent of the MgATP-dependent proton production was found to be determined by the redox state of the Fe protein: when the Fe protein was oxidized (and no electrons are transferred but MgATP is still hydrolysed) more protons were released ( k_obs ≈6 s ^-1) than in the case when the Fe protein was reduced. This was explained as an indication that during electron transfer, together with the electron, a proton is absorbed by the MoFe protein. Proton production was also observed when the nitrogenase proteins were mixed with MgADP; the characteristics of the proton production, however, differed from the MgATP-induced proton production. It was argued correctly that pH changes during nitrogenase turnover cannot unambiguously be assigned to a particular event in the nitrogenase catalytic cycle (Lowe et al., 1995): protons might be released due to the binding of MgATP to the nitrogenase complex, MgATP hydrolysis, the release of Pi, the release of MgADP, or during all of these steps. Lowe et al. (1995) studied the MgATP-induced pre-steady-state Pi release by Klebsiella pneumoniae nitrogenase, using a fluorescent phosphate probe. The reported Pi release curves are remarkably similar to our proton production curves. To be able to simulate their data, Lowe et al. (1995) had to adjust the Fe protein cycle (Chapter 1, Scheme 1) and use a kinetic scheme in which electron transfer ( k_obs= 176  s ^-1) precedes the hydrolysis of MgATP ( k_obs= 50  s ^-1) and the release of Pi from the nitrogenase complex ( k_obs= 22  s ^-1) As proposed in Chapter 2, it is likely that the binding of MgATP to the nitrogenase complex and a change of the conformation of the nitrogenase complex from an "as- isolated" to a "MgATP-bound" conformation, are the trigger for electron transfer from the Fe protein to the MoFe protein. A recent interesting finding in this respect is that deletion of Leu ¹²⁷from the nucleotide binding region of the (Azotobacter vinelandii) Fe protein results in a conformation which closely resembles the MgATP-bound state (Ryle & Seefeldt, 1996), and that this altered Fe protein is capable of transferring one electron to the MoFe protein in the absence of MgATP (Lanzilotta et al., 1996).

When the crystallographic structure of the Fe protein from A. vinelandii was solved, it was found that the core of the Fe protein and specifically the nucleotide binding site, is highly similar to the nucleotide binding site of the H-Ras p21 protein (Georgiadis et al., 1992). It was suggested that, like this protein and other molecular switch proteins, nitrogenase could make use of nucleotide binding and hydrolysis to switch between different protein conformations. Non- hydrolysable analogues of ATP or GTP are often used to study the mechanism of nucleotide hydrolysis. Aluminium fluoride is known to act as an analogue of phosphate in the presence of ADP/GDP for various ATPases and GTPases (Chabre, 1990). In Chapter 3 the formation of a stable complex between both nitrogenase proteins, MgADP and aluminium fluoride is described. This complex did not have nitrogenase activity. However, dissociation of the nitrogenase.ADP. aluminium fluoride complex could be stimulated by incubation of the complex at 50°C in the presence of P _i and NaCI, and nitrogenase activity was recovered. The composition of the nitrogenase.ADP. aluminium fluoride complex was ~2.7 Av2 and -2.0 ADP per Av1. EPR measurements showed that in the nitrogenase ADP- aluminium fluoride complex the MoFe protein was present in the as-isolated, dithionite-reduced state, whereas the Fe protein was oxidized. The crystal structures of molecular switch proteins complexed with aluminium fluoride and ADP/GDP (transducin αmyosin and G _iα revealed that ADP/GDP. aluminium fluoride mimics a transition state of ATP/GTP hydrolysis. Analogous to these findings, it is likely that we isolated the nitrogenase complex in a conformation which resembles a transition state in MgATP hydrolysis. Shortly after the publication of this work, Renner & Howard published a related study with similar results (1996). It is important to note that the complex with aluminium fluoride and ADP contains both the nitrogenase Fe protein and the MoFe protein, whereas in the case of the molecular switch proteins the nucleotide binding protein alone tightly binds aluminium fluoride and ADP/GDP. Exceptions are elongation factor EF-G, which forms a stable complex with aluminium fluoride, GDP and ribosomes but is not affected by aluminium fluoride in the absence of ribosomes (Mesters et al., 1993), and the Ras-protein, which only forms a complex with aluminium fluoride and GDP in the presence of certain GTPase activating proteins (Mittal et al., 1996). Recently the nitrogenase.ADP. aluminium fluoride complex was crystallized (H. Schindelin, C. Kisker, J. Howard & D. Rees, personal communication to Dr. H. Haaker). Preliminary data showed that the binding of aluminium fluoride and MgADP causes major conformational changes; in the nitrogenase.ADP -aluminium fluoride complex the subunits of the Fe protein are twisted and have moved ~20 A towards each other, MgADP is bound in the H-Ras-like orientation (along the subunit interface) and the "top" helices in the environment of the [4Fe4S] cluster are flat, which diminishes the distance between the [4Fe4S] cluster and the P-cluster by -5 Å compared to the distance in the proposed docking model for the nitrogenase complex (Kim & Rees, 1992b).

The pre-steady-state electron transfer reactions of nitrogenase are accompanied by changes of the absorbance at 430 nm. The interpretation of these absorbance changes is the subject of Chapter 4. The absorbance increase observed immediately after mixing of the nitrogenase proteins with MgATP is associated with the transfer of the first electron from the Fe protein to the MoFe protein and is mainly due to the oxidation of the Fe protein. After the initial absorbance increase (~50 ms) the absorbance decreases, which, dependent on the ratio [Av2]/[Av1], is followed by another absorbance increase. In Chapter 6 it is shown that this final absorbance increase is largely determined by the redox state of the Fe protein. Lowe et al. (1993) simulated the absorbance changes associated with the pre-steady-state electron transfer reactions of nitrogenase from K. pneunioniae, by ascribing the absorbance changes after the initial absorbance increase to successive redox changes of the MoFe protein (see Chapter 1, Scheme 2). The absorbance curves obtained for the nitrogenase proteins from A. vinelandii had a much more pronounced shape than the curves reported for K . pneumoniae nitrogenase (which did not contain a clear absorbance decrease) and, consequently, we were unable to simulate the A. vinelandii data with the model used by Lowe et al. (1993). We had to add an extra step to the kinetic scheme, with a rate constant of -14 s ^-1, to obtain an adequate simulation of the absorbance curves. When the reductant-free nitrogenase proteins (oxidized Av2) were mixed with MgATP, the absorbance at 430 nm decreased after a delay of -20 ms, with k_obs = 6.6 s ^-1; it was shown that the absorbance had shifted from 430 nm to 360 nm. It is highly probable that this absorbance decrease is associated with the reductant-independent ATPase activity of nitrogenase. The rate of this absorbance decrease and the rate of the extra reaction added to the kinetic scheme, were of the same order of magnitude as the rates of the MgATP-dependent pre-steady-state proton production in the absence ( k_obs ≈ 6 s ^-1) and in the presence of reductant ( k_obs ≈ 14 s ^-1) described in Chapter 2. It is proposed that the decrease of the absorbance at 430 nm observed during electron transfer and in the absence of reductant, is caused by a change of the conformation of the nitrogenase complex as a consequence of the hydrolysis of MgATP. This change from the MgATP-bound to the "MgADP-bound" conformation of the nitrogenase complex is accompanied by a release of protons.

A high concentration of NaCI and a low reaction temperature both lower the amplitude and the observed rate constant of the initial absorbance increase (430 nm) associated with pre-steady- state electron transfer from the Fe protein to the MoFe protein (Chapter 5). This suggests that, under such conditions, only a part of all MoFe protein present is reduced. Rapid-freeze EPR experiments showed that at 5°C (without NaCl) the reduction of the FeMo cofactor of the MoFe protein was indeed incomplete. This effect can be explained by assuming that the electron transfer between the nitrogenase proteins is a reversible process of which the back reaction becomes significant at low temperatures (Thorneley et al., 1989; Mensink & Maker, 1992). In the presence of 500 mM NaCI however, the incomplete reduction of the MoFe protein (~35 %) as suggested by the amplitude of the stopped-flow signal, was not confirmed by the rapidfreeze EPR data: ~85 % of FeMoco was found to be reduced. We concluded that reversibility of electron transfer could not account for the diminished absorbance amplitude observed in the presence of NaCl. It is proposed that NaCl inhibits the rate of electron transfer, but not the rate of MgATP hydrolysis and the subsequent conformational change of the nitrogenase complex. Because the change from the MgATP-bound to the MgADP-bound nitrogenase conformation is accompanied by an absorbance decrease (Chapter 4), the absorbance increase associated with electron transfer is overtaken by this subsequent absorbance decrease and does not reach the maximum value. Deits & Howard (1990) concluded from steady-state experiments that NaCI inhibits substrate reduction without altering the ratio MgATP hydrolysed/electrons transferred. This agrees with our conclusion that electron transfer occurs when the nitrogenase complex is in the MgATP-bound conformation and not after MgATP hydrolysis (Chapter 2 and Chapter 6). It was observed that in the presence of salt the reduction of FeMoco took place after a lag phase, whereas the absorbance increase associated with electron transfer started immediately after mixing of the nitrogenase proteins with MgATP. We propose that the electron transfer reaction is not a one-step process, but occurs from the Fe protein to FeMoco via an as-yet unidentified site at the MoFe protein. This site might be the P-cluster (Peters et al., 1995), but we were not able to find experimental evidence confirming this suggestion.

The dissociation of the nitrogenase complex after electron transfer and MgATP hydrolysis, is generally considered to be the rate-limiting step of the Fe protein cycle (Chapter 1, Scheme 1) and is thought to be obligatory for the re-reduction of the Fe protein and the exchange of MgADP for MgATP. This hypothesis was based on kinetic experiments with sodium dithionite as reductant. In Chapter 6 the rate-limiting step of the Fe protein cycle was studied, using three different reductants: sodium dithionite, Ti(III) citrate and flavodoxin hydroquinone. The rate of the dissociation of the nitrogenase complex was determined from the rate of reduction of oxidized Fe protein with MgADP bound (Av2 _ox (MgADP) ₂ ), in the presence of MoFe protein, by dithionite Robs = 2.8 s ^-1). Ti(III) citrate and flavodoxin hydroquinone rapidly reduced Av2 _ox (MgADP) ₂ in the nitrogenase complex, without preceding dissociation of the complex. The observation that dissociation of the nitrogenase complex is the ratelimiting step of nitrogenase catalysis (Thorneley & Lowe, 1983) is probably caused rather by the low concentration of the actual reductant, S0 ₂^-1, when sodium dithionite is used - which enables the nitrogenase complex to dissociate before Av2 _ox (MgADP) ₂ is reduced - rather than by an impossibility of reduction of Av2 _ox (MgADP) ₂ in the nitrogenase complex. Our data indicate that the exchange of MgADP for MgATP does not require dissociation of the reduced nitrogenase complex (Avl1.Av2 _red (MgADP) ₂ ) either. The uptake of electrons by nitrogenase during turnover was studied by monitoring the absorbance changes associated with the oxidation of Ti(III) citrate and flavodoxin hydroquinone. The presence of a slow phase in the pre-steady-state electron uptake curves indicated that a slow step occurs in the Fe protein cycle after reduction of the oxidized Fe protein. Without adding such a slow reaction ( k = 11 s ^-1) to the kinetic scheme, simulations of the kinetic cycle with Ti(III) citrate or flavodoxin hydroquinone as the reductant yielded a turnover rate which was unreasonably high compared to the observed rate of nitrogenase turnover. It is proposed that this slow step and the conformational change of the nitrogenase complex after MgATP hydrolysis (14 s ^-1) together limit the rate of nitrogenase turnover.

The sequence of events during nitrogenase catalysis: a revision of the Fe protein cycle

With the kinetic data described in this thesis it is possible to revise the Fe protein cycle (Chapter 1, Scheme 1) to a version (Chapter 6, Scheme 2) which, according to our kinetic data, better describes the sequence of events during nitrogenase catalysis than the original Fe protein cycle. The binding of MgATP to the nitrogenase complex induces a change from the "resting" conformation to the MgATP-bound conformation of the complex. This conformational change triggers electron transfer from the Fe protein to the MoFe protein ( k_obs ≈ 100 s ^-1) In the MgATP-bound conformation MgATP is hydrolysed ( k ≈ 77 s ^-1), which is followed by a change of the conformation of the nitrogenase complex from the MgATP-bound to the MgADP-bound conformation. This conformational change is accompanied by the production of protons and a decrease of the absorbance at 430 nm ( k_obs ≈ 14 s ^-1). A similar kinetic scheme was used by Lowe et al. (1995) to describe the pre-steady-state P, release by K. pneumoniae nitrogenase. The oxidized Fe protein is rapidly reduced in the nitrogenase complex if Ti(III) citrate or flavodoxin hydroquinone is used, and at a much lower rate if sodium dithionite is used. The slow step in the catalytic cycle might be a relaxation of the nitrogenase complex from the MgADP-bound conformation to the resting state ( k ≈ 11 s ^-1). After this step MgATP is exchanged for MgADP ( k_obs ≈ 55 s ^-1), and the nitrogenase complex enters the next round of electron transfer. In this scheme, dissociation of the nitrogenase complex is a side-reaction of the catalytic cycle and only occurs with sodium dithionite as reductant.

The role of MgATP in nitrogenase catalysis: the Fe protein as a molecular switch protein

When the close structural similarity between the nucleotide binding site of the Fe protein from A. vinelandii and the nucleotide binding site of the H-Ras p21 protein was discovered, it was suggested that the Fe protein might act as a molecular switch protein (Georgiadis et al., 1992). It has become clear that the resemblance between the nitrogenase Fe protein and the molecular switch proteins is more than just structural. We have presented kinetic evidence that conformational changes of the Fe protein and thus of the nitrogenase complex, induced by the binding of MgATP and MgATP-hydrolysis, drive the catalytic cycle of nitrogenase. That the MgATP-bound conformation of the Fe protein, as proposed in Chapter 1, is sufficient to induce electron transfer to the MoFe protein, was beautifully illustrated by Ryle & Seefeldt (1996) and Lanzilotta et al. (1996), who constructed an Fe protein capable of electron transfer in the absence of MgATP. As with many molecular switch proteins, aluminium fluoride and MgADP stabilize the nitrogenase complex in what is probably a transition state of MgATP hydrolysis, thereby inhibiting all nitrogenase activity. The MoFe protein induces MgATP hydrolysis at the Fe protein, which is reminiscent of the way an effector protein induces nucleotide hydrolysis by molecular switch proteins. The function of MgATP hydrolysis might be to secure such a conformation of the nitrogenase complex (the MgADP-bound conformation) that backflow of electrons from the MoFe protein to the Fe protein is prevented (Wolle et al., 1992). The resemblances between the structures and the catalytic mechanisms lead to the conclusion that the nitrogenase Fe protein must be regarded as a molecular switch protein.