Characterization of Azotobacter vinelandii nifZ deletion strains. Indication of stepwise MoFe protein assembly.

The nifZ gene product (NifZ) of Azotobacter vinelandii has been implicated in MoFe protein maturation. However, its exact function in this process remains largely unknown. Here, we report a detailed biochemical/biophysical characterization of His-tagged MoFe proteins purified from A. vinelandii nifZ and nifZ/nifB deletion strains DJ1182 and YM6A (Delta nifZ and Delta nifZ Delta nifB MoFe proteins, respectively). Our data from EPR, metal, activity, and stability analyses indicate that one alpha beta subunit pair of the Delta nifZ MoFe protein contains a P cluster ([8Fe-7S]) and an iron-molybdenum cofactor (FeMoco) ([Mo-7Fe-9S-X-homocitrate]), whereas the other contains a presumed P cluster precursor, possibly comprising a pair of [4Fe-4S]-like clusters, and a vacant FeMoco site. Likewise, the Delta nifZ Delta nifB MoFe protein has the same composition as the Delta nifZ MoFe protein except for the absence of FeMoco, an effect caused by the deletion of the nifB gene. These results suggest that the MoFe protein is likely assembled stepwise, i.e. one alpha beta subunit pair of the tetrameric MoFe protein is assembled prior to the other, and that NifZ might act as a chaperone in the assembly of the second alpha beta subunit pair by facilitating a conformational rearrangement that is required for the formation of the P cluster through the condensation of two [4Fe-4S]-like clusters. The possibility of NifZ exercising its effect through the Fe protein was ruled out because the Fe proteins from nifZ and nifZ/nifB deletion strains are not defective in their normal functions. However, the detailed mechanism of how NifZ carries out its exact function in MoFe protein maturation awaits further investigation.

The biochemical machinery for the reduction of dinitrogen to ammonia is provided by the metalloenzyme nitrogenase. This enzyme comprises two separately purifiable proteins, the MoFe (molybdenum-iron) protein and the Fe (iron) protein (reviewed in Refs. 1-7). 1 The Fe protein is a 60-kDa dimer of two identical subunits encoded by the nifH gene. Each subunit of the Fe protein contains a binding site for MgATP, and the two subunits are bridged by a single [4Fe-4S] cluster. The more complicated MoFe protein is an ␣ 2 ␤ 2 tetramer of ϳ230 kDa, and its ␣ and ␤ subunits are encoded by the nifD and nifK genes, respectively. The MoFe protein contains two different types of unique metal clusters, i.e. the P cluster ([8Fe-7S]), which is bridged between each ␣␤ subunit pair, and the iron-molybdenum cofactor (FeMoco) 2 ([Mo-7Fe-9S-X-homocitrate]), 3 which is located within each ␣ subunit.
The substrate reduction activity of nitrogenase requires the participation of both component proteins, with the Fe protein serving as an obligate electron donor to the MoFe protein, which in turn provides the site of substrate reduction. The reduced Fe protein first binds two molecules of MgATP and undergoes a conformational change before forming a complex with the MoFe protein. Then, coupled with the hydrolysis of MgATP, electrons are transferred from the [4Fe-4S] cluster of the Fe protein to the P cluster of the MoFe protein within the complex. This process is followed by complex dissociation, rereduction of the oxidized Fe protein, and dissociation of MgADP, allowing the enzyme complex to start the next cycle of electron transfer. Finally, electrons are believed to be transferred from the P cluster to FeMoco, where substrate binding and reduction take place.
The assembly of nitrogenase and, in particular, the MoFe protein and its two unique metalloclusters is arguably one of the most complicated processes in the field of bioinorganic chemistry (reviewed in Refs. 9 -13, 15-19, and 21). Based on the current model, the assembly of the MoFe protein requires the participation of at least 15 different genes (9 -13, 15) and involves the following events: (i) the biosynthesis of FeMoco, (ii) the biosynthesis of a P cluster-containing yet FeMoco-deficient species of the MoFe protein in a separate pathway, and (iii) the insertion of completed FeMoco into the FeMoco-deficient MoFe protein (3, 9, 16 -19). It has been proposed that FeMoco biosynthesis starts with the mobilization of iron and sulfur and the assembly of Fe-S fragments that are subsequently transferred to NifB (nifB gene product), where the Fe-S core of FeMoco (designated NifB-co) is formed. NifB-co is then transferred to NifEN (nifE and nifN gene products), where it rearranges in a reaction that probably requires the Fe protein (nifH gene product). Molybdenum and homocitrate may or may not be inserted at this point, but they are attached prior to the insertion of FeMoco into a P cluster-containing but FeMoco-deficient form of the MoFe protein during the final maturation of the holo-MoFe protein. Such a form of MoFe protein was isolated from a nifB deletion strain of Azotobacter vinelandii (designated the ⌬nifB MoFe protein). It can be directly activated by isolated FeMoco, indicating that the FeMoco-binding site in the ⌬nifB MoFe protein is readily accessible for FeMoco insertion (22)(23)(24)(25)(26). This is consistent with the crystal structure of the ⌬nifB MoFe protein, which is essentially unchanged from that of the wild-type protein except for the rearrangement of a domain presumed to facilitate the insertion of FeMoco into the accessible FeMoco-binding site in the MoFe protein (27).
Contrary to the progress that has been made regarding the biosynthesis and insertion of FeMoco, the assembly of the P cluster remains largely unknown. Recently, we were able to show that a MoFe protein isolated from a nifH deletion strain of A. vinelandii (designated the ⌬nifH MoFe protein) does not contain a fully assembled P cluster; rather, it has a P cluster precursor that is composed of two [4Fe-4S]-like clusters (28). This is consistent with our earlier biochemical and spectroscopic studies indicating the presence of an "unusual" P cluster in this ⌬nifH MoFe protein (29). These studies provide the first insights into the process of P cluster formation, which occurs likely by the condensation of two [4Fe-4S]-like fragments, possibly concomitant with Fe protein (nifH gene product)-induced conformational change (28).
Although a number of components have been implicated in the complex assembly process of the MoFe protein and its clusters, more players are yet to be identified before a more accurate model of nitrogenase assembly can be established. It has been demonstrated that NifZ (nifZ gene product) is required for the expression of a fully functional MoFe protein in A. vinelandii (30). Some earlier studies suggest that NifZ has some function related to formation of FeMoco or insertion of FeMoco into the immature MoFe protein (30), whereas more recent reports indicate that it might be involved in P cluster synthesis (31,32). Here, we report a detailed biochemical/ biophysical characterization of MoFe proteins expressed by the nifZ and nifZ/nifB deletion strains of A. vinelandii and the implication of these studies for the function of NifZ and the assembly mechanism of the MoFe protein.

Materials
Unless noted otherwise, all chemicals and reagents were obtained from Fisher, Baxter Scientific, or Sigma. Construction of nifZ Deletion Strain AvYM3A-We constructed plasmid pHR5, which contains a 225-bp in-frame deletion in the nifZ gene that results in a deletion between the 46th and 120th codons of the nifZ gene product. This plasmid was then transformed into the wild-type A. vinelandii strain (AvOP) using a previously described method (34), resulting in strain YM3A, which expresses a non-His-tagged MoFe protein and has a 225-bp in-frame deletion of nifZ.

Construction of Variant A. vinelandii Strains
Construction of nifZ Deletion Strain AvDJ1182-Two plasmids were constructed: pDB131, which has a 1.3-kb kanamycin resistance cartridge inserted into the nifZ gene at the PstI site, and pDB264, which contains a 99-bp Bal-31-generated in-frame deletion in nifZ that originates at the PstI site, resulting in a deletion between the 48th and 80th codons of the nifZ gene product. Plasmid pDB131 was transformed into AvDJ1141 (22), an A. vinelandii strain expressing an N-terminally His-tagged MoFe protein. The resulting kanamycin-resistant strain is AvDJ1169, which expresses an N-terminally His-tagged MoFe protein and has a kanamycin resistance cartridge inserted into the nifZ gene at the PstI site. Plasmid pDB264 was then transformed into AvDJ1169.
The resulting kanamycin-sensitive strain is AvDJ1182, which expresses an N-terminally His-tagged MoFe protein and has a 99-bp in-frame deletion in nifZ. Construction of nifZ/nifB Deletion strain AvYM6A-Following the methods described previously (22), we constructed plasmid pHR18, which has an internal portion of nifB (flanked by SphI sites) removed and replaced by a 1.3-kb kanamycin resistance cartridge. This plasmid was then transformed into AvDJ1182. The resulting kanamycin-resistant strain is AvYM6A, which expresses an N-terminally His-tagged MoFe protein and has a 99-bp in-frame deletion in nifZ plus a deletion/ kanamycin resistance cartridge insertion in nifB.

Cell Growth and Protein Purification
Wild-type and variant strains of A. vinelandii were grown in 180-liter batches in a 200-liter New Brunswick fermentor in Burke's minimal medium supplemented with 2 mM ammonium acetate. The growth rate was measured by cell density at 436 nm using a Spectronic 20 Genesys (Spectronic Instruments, Rochester, New York). After the consumption of the ammonia, the cells were de-repressed for 3 h, followed by harvesting using a flow-through centrifugal harvester (Cepa). The cell paste was washed with 50 mM Tris-HCl (pH 8.0).

EPR Spectroscopy
All EPR samples were prepared in a Vacuum Atmospheres dry box with an oxygen level of Ͻ4 ppm. Unless noted otherwise, all samples were in 25 mM Tris-HCl (pH 8.0), 10% glycerol, and 2 mM Na 2 S 2 O 4 . MoFe protein samples were oxidized by incubation with excess indigo disulfonate (IDS) for 30 min. Subsequently, IDS was removed by a single pass over an anion exchange column as described (37). Samples were either used as they were or concentrated in a Centricon-30 (Amicon, Inc.) in anaerobic centrifuge tubes outside of the dry box. All perpendicular and parallel mode EPR spectra were recorded using a Bruker ESP 300 E z spectrophotometer interfaced with an Oxford Instruments ESR-9002 liquid helium continuous flow cryostat. All spectra were recorded at 10 K using a microwave power of 50 milliwatts, a gain of 5 ϫ 10 4 , a modulation frequency of 100 kHz, and a modulation amplitude of 5 G. Microwave frequencies of 9.62 and 9.39 GHz were used for the perpendicular (10 scans) and parallel (20 scans) mode EPR spectra, respectively.

Spin Quantitation of EPR Signals
Spin quantitation of the novel S ϭ 1/2 perpendicular mode EPR signal in the MoFe protein variant was carried out under non-saturating conditions with Cu 2ϩ -EDTA as the standard following the proce-  (38). This method has been applied to and outlined for quantitation of similar spin systems in earlier nitrogenase-and ferredoxin-related studies (26, 29, 39 -42). Quantitation numbers in this study represent the average of four independently purified protein batches and show an uncertainty of 3.5%. Spin quantitation of the S ϭ 3/2 perpendicular mode EPR signal was accomplished by comparison of the integrated area of the FeMoco signal in the MoFe protein variant with that in the wild-type protein. It has been well established that the S ϭ 3/2 EPR signal of the wild-type MoFe protein originates from the FeMoco center of the protein and integrates to 1 spin/FeMoco (17,44). The wild-type MoFe protein used in this study contains a complete cluster content based on metal analysis and activity measurements and can therefore be used as an internal standard for spin quantitation. This method has been used in earlier MoFe proteinrelated studies involving catalytic and structural investigations (36,45,46). Quantitation numbers in this study represent the average of three independently purified protein batches and show an uncertainty of 2.9%.
Spin quantitation of the g ϭ 11.8 parallel mode EPR signal was performed using the wild-type MoFe protein as an internal standard. The wild-type MoFe protein has been successfully applied to earlier studies as a standard (22,47), and quantitation numbers in this study represent the average of three independently purified protein batches and show an uncertainty of 5.6%. It should be noted that the results of the spin quantitation of the MoFe protein variants (S ϭ 3/2 perpendicular mode EPR signal and g ϭ 11.8 parallel mode EPR signal) are also consistent with their specific activities determined elsewhere in this study.

Activity Assays and Metal Analysis
All nitrogenase activity assays (8.7-ml vials) were carried out as described previously (33). The products H 2 and C 2 H 4 were analyzed as described (48). Ammonium was determined by a high performance liquid chromatography fluorescence method (49). Assays to determine the maximal activation of FeMoco-deficient MoFe proteins by insertion of isolated FeMoco in N-methylformamide (50) and those to test the ability of the Fe proteins to participate in FeMoco biosynthesis and insertion (36) were performed as described previously. Molybdenum (51) and iron (52) were determined as reported previously.

Protein Stability Experiments
Two approaches were used to determine the stability of the purified MoFe proteins: gel filtration and heat treatment. For gel filtration, a total amount of 100 mg of purified MoFe protein (in 25 mM Tris-HCl (pH 8.0), 10% glycerol, 250 mM NaCl, and 2 mM Na 2 S 2 O 4 ) was loaded onto an AcA34 gel filtration column (2.5 ϫ 130 cm, 600-ml bed volume; ICF, Cedex, France) and chromatographed as described (29). For heat treatment, a total amount of 50 mg of purified MoFe protein (in the same buffer described above) was incubated in a crimped anaerobic vial (8.7-ml volume; 100% argon gas atmosphere) at 56°C for 30 s. Precipitated protein was subsequently removed in both cases by centrifugation at 10,000 rpm for 10 min in a Biofuge fresco (Heraeus) in a Vacuum Atmospheres dry box. The enzymatic activity of the MoFe protein after gel filtration and heat treatment was determined as described previously (33). The product of each assay was then analyzed as reported (48).

Growth of nifZ and nifZ/nifB Deletion Strains of A. vinelandii under N 2 Fixation Conditions
The wild-type strains AvOP and AvDJ1141 showed nearly identical cell growth under N 2 fixation conditions ( Fig. 1), indicating that the His-tagged wild-type MoFe protein (expressed in AvDJ1141) is not appreciably different from the normal wild-type MoFe protein (expressed in AvOP) in terms of catalytic properties (22). Consistent with this observation, nifZ deletion strains expressing the non-His-and His-tagged MoFe proteins (AvYM3A and AvDJ1182, respectively) also showed nearly identical cell growth under N 2 fixation conditions ( Fig.  1). However, in contrast to AvOP and AvDJ1141, their growth rates decreased significantly. The doubling time of AvYM3A or AvDJ1182 under N 2 fixation conditions was ϳ7 h compared with ϳ4 h in the case of AvOP or AvDJ1141. The decreased growth rate of these nifZ deletion strains indicates a reduced substrate-reducing ability of ⌬nifZ MoFe proteins. This is in line with earlier crude extract experiments suggesting that the nifZ gene product is required for full expression, stability, or processing of the MoFe protein (30).
As expected, the nifZ/nifB deletion strain AvYM6A (expressing the His-tagged ⌬nifZ⌬nifB MoFe protein) was not able to grow under N 2 fixation conditions (Fig. 1). This is a result of the deletion of the nifB gene in the strain, which leads to the expression of an FeMoco-deficient inactive MoFe protein (22,27).

Purification of ⌬nifZ and ⌬nifZ⌬nifB MoFe Proteins
Using the previously described methods (26,29,36), up to 400 mg of the non-His-or His-tagged ⌬nifZ MoFe protein was purified from 200 g AvYM3A or AvDJ1182 cells, respectively, whereas up to 200 mg of the His-tagged ⌬nifZ⌬nifB MoFe protein was purified from 200 g AvYM6A cells. As shown in Fig.  2 (A, lane 3; and B, lanes 3 and 4), the non-His-and His-tagged ⌬nifZ MoFe proteins as well as the His-tagged ⌬nifZ⌬nifB MoFe protein are all composed of ␣ and ␤ subunits. The molecular mass of these MoFe protein variants are identical to that of the wild-type protein based on their elution profiles on a Sephacryl S-200 HR gel filtration column (Amersham Biosciences) (data not shown), indicating that these MoFe protein variants are ␣ 2 ␤ 2 tetramers with a molecular mass of ϳ230,000 Da.

EPR Spectroscopic Properties of ⌬nifZ and ⌬nifZ⌬nifB
MoFe Proteins Perpendicular Mode EPR Measurements-The isolated wildtype MoFe protein (expressed in AvOP) exhibits a well characterized S ϭ 3/2 EPR signal (Fig. 3, trace 1) that arises from the FeMoco center of the protein (1,17). The same S ϭ 3/2 EPR signal (Fig. 3, trace 3) was observed in the case of the isolated His-tagged ⌬nifZ MoFe protein (expressed in AvDJ1182), which integrates to 54% of that of the wild-type protein, indicating the presence of ϳ50% of wild-type FeMoco content in this MoFe protein variant. In addition, in contrast to the wildtype MoFe protein (Fig. 3, trace 1), the His-tagged ⌬nifZ MoFe protein exhibits an additional S ϭ 1/2 EPR signal in the g Ϸ 2 region, which integrates to 2 spins/MoFe protein (trace 3). The same signal has been observed in the case of the ⌬nifH MoFe protein (29), which was recently attributed to the presence of a presumed P cluster precursor in the protein that is composed of two [4Fe-4S]-like clusters (28).
Although the same S ϭ 3/2 and S ϭ 1/2 EPR signals (Fig. 3, trace 2) were observed in the isolated non-His-tagged ⌬nifZ MoFe protein (expressed in AvYM3A), the intensities of these signals were significantly weaker than those of the His-tagged ⌬nifZ MoFe protein (trace 3). The S ϭ 3/2 and S ϭ 1/2 signals of the non-His-tagged ⌬nifZ MoFe protein integrate to 22% of that of the wild-type protein and ϳ0.6 spins/protein, respectively, in comparison with 54% and ϳ2 spins/protein, respec-tively, in the His-tagged protein. The loss of these spectroscopic properties of the non-His-tagged ⌬nifZ MoFe protein can be explained by the destruction of the fragile protein and the loss of its containing metal clusters during the time-consuming purification procedure, an effect greatly minimized in the case of the His-tagged ⌬nifZ MoFe protein, which could be isolated by an efficient one-step purification procedure. The same effect has been reported previously in the case of the fragile ⌬nifH MoFe protein when the non-His-and His-tagged versions of this protein are compared (29). In this case, the fast purification procedure of the His-tagged ⌬nifH MoFe protein allowed the identification of an unstable P cluster precursor, which was impossible in the case of the non-His-tagged ⌬nifH MoFe protein (28). Because of these observations, we performed our further characterization of the ⌬nifZ MoFe protein using the strain expressing the His-tagged version of the protein (i.e. AvDJ1182), as will be discussed below.
The isolated His-tagged ⌬nifZ⌬nifB MoFe protein (expressed in AvYM6A) does not exhibit the S ϭ 3/2 EPR signal (Fig. 3, trace 4) arising from the FeMoco center in the wild-type and ⌬nifZ MoFe proteins (traces 1-3). This result is to be expected because AvYM6A contains a deletion in nifB that results in the expression of an FeMoco-deficient form of the MoFe protein in this strain. However, like the His-tagged ⌬nifZ MoFe protein (Fig. 3, trace 3), the His-tagged ⌬nifZ⌬nifB MoFe protein exhibits an S ϭ 1/2 EPR signal in the g Ϸ 2 region, which integrates to ϳ2 spins/MoFe protein (trace 4), suggesting the possible presence of a P cluster precursor in both MoFe protein variants similar to that described in the case of the ⌬nifH MoFe protein (28,29).
Parallel Mode EPR Measurements-It has been established that upon IDS oxidation, the P N state of the P cluster, which is present in the isolated wild-type and ⌬nifB MoFe proteins, can be two-electron-oxidized to the P 2ϩ state (22), which can then be recognized by a g ϭ 11.8 parallel mode EPR signal (22,47,53). The IDS-oxidized His-tagged ⌬nifZ MoFe protein exhibits this g ϭ 11.8 signal (Fig. 4, trace 2), which integrates to 51% of that of the wild-type protein (trace 1), indicating the presence of ϳ50% of wild-type P cluster content in this MoFe protein variant. Meanwhile, as shown in Fig. 4 (trace 3), the IDS-oxidized His-tagged ⌬nifZ⌬nifB MoFe protein exhibits a g ϭ 11.8 par- allel mode EPR signal of the same intensity as that of the IDS-oxidized His-tagged ⌬nifZ MoFe protein (trace 2), suggesting the presence of the same amount of fully assembled P cluster (ϳ50% of wild-type P cluster content) in both MoFe protein variants. The fact that the respective S ϭ 1/2 perpendicular mode (Fig. 3, traces 3 and 4) and g ϭ 11.8 parallel mode (Fig. 4, traces 2 and 3) EPR signals observed in His-tagged ⌬nifZ and ⌬nifZ⌬nifB MoFe proteins are of nearly identical intensity is also consistent with previous reports that NifB is specifically involved in the biosynthesis of FeMoco and does not affect the production of the P cluster (27).

Metal Analysis of ⌬nifZ and ⌬nifZ⌬nifB MoFe Proteins
Based on the EPR spectroscopy experiments above, the Histagged ⌬nifZ MoFe protein has half of the wild-type FeMoco and P cluster content, i.e. 1 molybdenum atom and 15 iron atoms per MoFe protein molecule. Metal analysis (Table II) showed that each molecule of His-tagged ⌬nifZ MoFe protein contains 1.15 Ϯ 0.02 molybdenum atoms and 22.6 Ϯ 0.5 iron atoms. If 1 molybdenum atom and 15 iron atoms are assigned to one FeMoco and one P cluster per molecule, then the rest of the iron atoms (ϳ8) could be assigned to one presumed P cluster precursor comprising two [4Fe-4S]-like clusters (possibly in the ϩ1 oxidation state), which give rise to a typical S ϭ 1/2 ground state EPR signal in the g Ϸ 2 region (54) that integrates to ϳ2 spins/molecule (Fig. 3, trace 3). As mentioned above, a similar P cluster precursor has been proposed in the case of ⌬nifH MoFe protein, which exhibits the same S ϭ 1/2 signal in the g Ϸ 2 region (28,29).
Consistent with the absence of FeMoco, the His-tagged ⌬nifZ⌬nifB MoFe protein does not contain detectable amounts of molybdenum (Table II). Nonetheless, parallel mode EPR spectroscopy indicated that this MoFe protein variant does have half of the wild-type P cluster content, i.e. 8 iron atoms/ MoFe protein molecule. Based on the metal analysis, each molecule of His-tagged ⌬nifZ⌬nifB MoFe protein contains 15.8 Ϯ 0.2 iron atoms (Table II). If 8 iron atoms are assigned to one P cluster, the other 8 iron atoms in the His-tagged ⌬nifZ⌬nifB MoFe protein, like those in the His-tagged ⌬nifZ MoFe protein, can again be assigned to one presumed P cluster precursor comprising two [4Fe-4S]-like clusters, which accounts for the S ϭ 1/2 EPR signal that integrates to ϳ2 spins/ molecule (Fig. 3, trace 4).

Activity Assays of ⌬nifZ and ⌬nifZ⌬nifB MoFe Proteins
Nitrogenase combines the reduction of N 2 to ammonia with the concomitant production of H 2 (55) 4 in a reaction that is commonly referred to as nitrogen fixation. Besides its physiological substrate N 2 , nitrogenase is also able to reduce a large variety of small double-and triple-bonded substrates (1,3,55,57), and the reduction of these alternative substrates by nitrogenase can be used to determine the activity of this enzyme. One of the most commonly used alternative substrates in nitrogenase activity that is acetylene (C 2 H 2 ), which is reduced to ethylene (C 2 H 4 ) that is easily detectable by gas chromatography analysis (1). Meanwhile, in the absence of any reducible substrates (e.g. in a gas atmosphere of 100% argon), nitrogenase catalyzes H 2 evolution, which results from the reduction of protons (55). This reaction is also commonly included in standard activity assays of nitrogenase. The half FeMoco and P cluster content of the His-tagged ⌬nifZ MoFe protein, which is based on EPR experiments (Fig. 3, trace 3; and Fig. 4, trace 2), correlates well with the specific activities of this MoFe protein variant in terms of H 2 evolution, C 2 H 2 reduction, and N 2 fixation, which are all ϳ50% of those of the wild-type protein (Table  III).
Because of the absence of FeMoco (Fig. 3, trace 4), the Histagged ⌬nifZ⌬nifB MoFe protein does not support substrate reduction activities (Table III). On the other hand, using the His-tagged ⌬nifB MoFe protein that forms fully active holo-MoFe protein upon the addition of isolated FeMoco (22, 27) as a standard, the His-tagged ⌬nifZ⌬nifB MoFe protein can be activated by FeMoco to ϳ50% in terms of H 2 evolution, C 2 H 2 reduction, and N 2 fixation activities (Table IV). This is consistent with the EPR data, which indicate that this MoFe protein variant contains 51% of the wild-type P cluster content (Fig. 4,  trace 3).
The 50% wild-type activity of the His-tagged ⌬nifZ MoFe protein could be explained only by the presence of one FeMoco and one P cluster in one ␣␤ subunit pair of the tetrameric protein, whereas the 50% activation of the His-tagged ⌬nifZ⌬nifB MoFe protein upon FeMoco addition could be interpreted only as one FeMoco being inserted into its binding site in the ␣␤ subunit pair that contains one fully assembled P cluster. This argument is based upon our previous studies of the ⌬nifH MoFe protein (28,29), which show that this P cluster 4 The overall reaction catalyzed by nitrogenase is usually depicted as follows (3,21,56):

Indication of Stepwise MoFe Protein Assembly
precursor-containing MoFe protein variant cannot be reconstituted by the addition of isolated FeMoco (Table IV). Meanwhile, although one FeMoco and P cluster pair should be assigned to one ␣␤ dimer of the His-tagged ⌬nifZ MoFe protein, the determination of one FeMoco, one P cluster, and one P cluster precursor per protein molecule reflects only an average cluster content of this MoFe protein variant. The question of whether the His-tagged ⌬nifZ MoFe protein is a homogeneous or mixed protein species was addressed by the following experiments.

Homogeneity of the His-tagged ⌬nifZ MoFe Protein
Based on the EPR experiments, metal analysis, and activity assays described above, the His-tagged ⌬nifZ MoFe protein could be either (i) a homogeneous species of MoFe protein containing one FeMoco, one P cluster, and one P cluster precursor comprising a pair of [4Fe-4S]-like clusters per molecule or (ii) a mixture of 50% wild-type MoFe protein containing two FeMoco and two P clusters per molecule and 50% ⌬nifH-type MoFe protein containing two P cluster precursors comprising two pairs of [4Fe-4S]-like clusters per molecule (28). Unfortunately, it was not possible to separate wild-type and ⌬nifH MoFe proteins by anion exchange or gel filtration chromatography (data not shown). In addition, the wild-type and all variant MoFe proteins in this study were unstable during mass spectroscopic experiments and degraded into equimolar amounts of ␣ and ␤ polypeptides (data not shown), making it impossible to directly assess the homogeneity of the protein samples. Therefore, an indirect approach based on the well known instability of FeMoco-deficient MoFe proteins and, in particular, the ⌬nifH MoFe protein upon heat or gel filtration chromatography treatment (26,29,36) was applied here. As shown in Table V, 5 the vast majority of the His-tagged wildtype MoFe protein could be recovered after heat or gel filtration chromatography treatment, and the specific activity of the protein remained essentially unchanged. In contrast to this, the His-tagged ⌬nifH MoFe protein appeared to be extremely unstable in both cases. Only 10 and 36% of the His-tagged ⌬nifH MoFe protein could be recovered after heat and gel filtration chromatography treatment, respectively (Table V). It is therefore to be expected that the majority of the recovered protein in a mixture of equal amounts of His-tagged wild-type and ⌬nifH MoFe proteins after heat or gel filtration chromatography treatment (42 and 58%, respectively) is the wild-type protein, which accounts for the increase in specific activity (90 and 45%, 5 Based on substrate reduction activities, metal contents, and EPR spectroscopic features (data not shown), the catalytic properties of the His-and non-His-tagged wild-type MoFe proteins in this study (expressed by AvDJ1141 and AvOP, respectively) are nearly identical. This observation is consistent with earlier reports (22), and for the purpose of simplicity, only the non-His-tagged wild-type MoFe protein was used as a standard in experiments involving both the His-and non-Histagged MoFe protein variants. Meanwhile, the His-tagged wild-type MoFe protein was used as a standard for the stability test (see Table V) because only the His-tagged MoFe protein variants were involved in this experiment (also see "Experimental Procedures"). However, it should be noted that the His-and non-His-tagged wild-type MoFe proteins showed nearly identical behavior in the stability experiment (data not shown).   respectively) in both cases (Table V). A completely different effect could be observed in the case of the His-tagged ⌬nifZ MoFe protein. The majority of protein could be recovered after heat or gel filtration chromatography treatment (98 and 81%, respectively), and the specific activities decreased (12 and 12%, respectively) in both cases (Table V). These results exclude the possibility that the His-tagged ⌬nifZ MoFe protein is a mixture of wild-type and ⌬nifH-type MoFe proteins; rather, it is a homogeneous species of MoFe protein containing one FeMoco, one P cluster, and one P cluster precursor comprising a pair of [4Fe-4S]-like clusters per molecule of protein (Fig. 5A). The results from two additional lines of experiments are consistent with the protein stability assays described above. (i) The Histagged ⌬nifH-type MoFe protein degraded during anaerobic isoelectric focusing and native polyacrylamide gel electrophoresis, producing a pattern of multiple bands in the stained gels. However, His-tagged wild-type and ⌬nifZ MoFe proteins appeared to be stable in these experiments, and only one protein band could be observed in both cases (data not shown  (Fig. 5B).

Catalytic and Biophysical Properties of the Fe Proteins
Expressed by AvYM3A, AvDJ1182, and AvYM6A The Fe protein of nitrogenase has at least three separate functions in the cell: (i) donation of electrons to the MoFe protein in support of N 2 reduction, (ii) participation in the initial biosynthesis of FeMoco, and (iii) insertion of preformed FeMoco into an FeMoco-deficient form of the MoFe protein (36). Therefore, the incomplete cluster content of the ⌬nifZ and ⌬nifZ⌬nifB MoFe proteins described above could be the result of an indirect effect of nifZ deletion leading to the expression of a defective Fe protein that is unable to function normally in MoFe protein maturation. To exclude this possibility, we per-formed detailed catalytic and biophysical characterization of the Fe proteins expressed by AvYM3A, AvDJ1182, and AvYM6A.
The molecular masses of the Fe proteins purified from the three variant A. vinelandii strains AvYM3A, AvDJ1182, and AvYM6A (Fig. 2C, lanes 3-5) appear to be identical to that of the wild-type Fe protein based on their elution profiles upon gel filtration chromatography (data not shown). All these Fe proteins contain the same amount of iron (ϳ4 iron atoms/molecule of Fe protein) as the wild-type protein (Table II). In addition, these Fe proteins exhibit an S ϭ 1/2 EPR signal in the g ϭ 2 region (Fig. 6, traces 2-4), which is qualitatively and quantitatively indistinguishable from that of the wild-type Fe protein (trace 1). These data suggest that the Fe protein of AvYM3A, AvDJ1182, or AvYM3A has a normal complement of the [4Fe-4S] cluster and that the cluster can be reduced by dithionite to an oxidation state of ϩ1. Consistent with this observation, all three Fe proteins have nearly the same activities as the wildtype protein in terms of H 2 evolution, C 2 H 2 reduction, and N 2 fixation (Table VI), indicating that they can function normally as electron donors for the MoFe protein in the substrate reduction reaction. Finally, these Fe proteins are indistinguishable from the wild-type protein in in vitro FeMoco biosynthesis and insertion assays (Table VII), further excluding the possibility that these Fe proteins are defective in any of their normal functions and that the incomplete cluster content of the ⌬nifZ and ⌬nifZ⌬nifB MoFe proteins results indirectly from the effect of nifZ deletion on the Fe protein.

Conclusion
Based on our study, we were able to show that the deletion of nifZ does not affect the formation of one-half of the MoFe protein comprising one ␣␤ subunit pair, a fully assembled P cluster bridged between the ␣ and ␤ subunits, and one FeMoco located at its binding site in the ␣ subunit. Such a fully assembled ␣␤ subunit pair of the MoFe protein can be viewed as a "catalytic unit" because it can carry out substrate reduction FIG. 5. Schematic diagrams of the ␣ 2 ␤ 2 tetrameric His-tagged ⌬nifZ MoFe protein (A) and His-tagged ⌬nifZ⌬nifB MoFe protein (B). A, the His-tagged ⌬nifZ MoFe protein is composed of one ␣␤ subunit pair that has one P cluster bridged between the ␣ and ␤ subunits and one FeMoco located in the ␣ subunit (a catalytic unit) and the other ␣␤ subunit pair that has a presumed P cluster precursor comprising a pair of [4Fe-4S]-like clusters located between the ␣ and ␤ subunits and a vacant FeMoco-binding site in the ␣ subunit. B, the His-tagged ⌬nifZ⌬nifB MoFe protein has the same composition as the His-tagged ⌬nifZ MoFe protein (A) except for the absence of FeMoco in both FeMoco-binding sites.  (trace 4). The protein concentration was 10 mg/ml. The spectra were measured as described under "Experimental Procedures." reactions with half of the activity of the ␣ 2 ␤ 2 tetramer regardless of the other half of the protein. On the other hand, the deletion of nifZ does have a major impact on the assembly of the second ␣␤ subunit pair of the MoFe protein, resulting in the presence of a presumed P cluster precursor, most likely composed of a pair of [4Fe-4S]-like clusters, and a vacant FeMocobinding site, which is inaccessible to FeMoco insertion, in the second half of the protein. These results suggest that MoFe protein assembly occurs in a stepwise fashion, i.e. one-half (or one catalytic unit) of the protein is assembled prior to the second half of the protein, and that NifZ participates in the assembly of the second half of the protein, in particular, the formation of the second P cluster. It is possible that, after the formation of one P cluster in one ␣␤ subunit pair, likely through condensation of two [4Fe-4S]-like clusters (28), the two [4Fe-4S]-like clusters in the other ␣␤ subunit pair are forced into a conformation that does not allow proper cluster condensation in the absence of NifZ. Consistent with this hypothesis, no spin coupling of the two [4Fe-4S]-like clusters could be observed upon EPR spectroscopy of the His-tagged ⌬nifZ or ⌬nifZ⌬nifB MoFe protein, indicating a distance of at least ϳ10 Å between the two clusters. NifZ might be required at this step of the assembly and acts in a chaperone-like fashion to facilitate a conformational rearrangement that places the two [4Fe-4S]-like clusters in proper distance and orientation for the condensation of the clusters into the P cluster (Fig. 7), and the deletion of nifZ "uncouples" the assembly of the two halves of the protein, resulting in a MoFe protein species that has the P cluster of the second ␣␤ subunit pair in the precursor state.
Our hypothesis of stepwise assembly of the MoFe protein that "distinguishes" the two apparently identical halves of the protein is further supported, in a somewhat indirect sense, by previous studies that demonstrate that the two ␣␤ subunit pairs of the MoFe protein can be different in terms of stability and cluster composition. An A. vinelandii strain expressing the E146D Fe protein variant, which is specifically defective in FeMoco insertion, accumulates a homogeneous MoFe protein species (designated the E146D nifH MoFe protein), which has one P cluster and one FeMoco in one-half of the molecule and one P cluster and one vacant FeMoco site in the other half (36,50). Purifications of the wild-type MoFe protein from Klebsiella pneumoniae usually yield a portion of MoFe protein with the same cluster content as the E146D nifH MoFe protein in addition to the fully assembled MoFe protein species (58). This effect has been attributed to the instability of the wild-type MoFe protein from K. pneumoniae and could be further explained by the different stability of the two ␣␤ subunit pairs of this protein. In addition to these MoFe protein species, an incomplete ␣␤ 2 form of the A. vinelandii VFe protein has been isolated, and this has been ascribed to ␣ and ␤ subunits being more loosely bound to each other in the VFe proteins compared with the MoFe proteins (20). In light of our findings, this could be further interpreted by the difference in the stability of the two incomplete ␣␤ halves of the VFe protein molecule that results in the loss of an ␣ subunit from one ␣␤ half. This may also reflect a common theme that the two halves of all MoFe proteins or MoFe protein equivalents of the alternative nitrogenase systems could have different properties, and further studies along the line of nifZ deletion strains may help to address questions regarding the stepwise assembly process that results in the "unequal" two halves of the protein molecule.
It should be noted that our results indicating a role of NifZ in P cluster biosynthesis appear somewhat contradictory to earlier crude extract experiments suggesting that the function of NifZ in MoFe protein maturation is likely related to an event involving or affecting FeMoco (30). However, the effect of NifZ on FeMoco insertion could be an indirect one resulting, from the effect of NifZ on P cluster formation and the conformational rearrangement associated with the process that affects the subsequent FeMoco insertion. Unfortunately, the nifZ gene does not show similarity to any sequences of the known proteins (based on a search of A. vinelandii nifZ sequence (Swiss Protein Database) using the program BLAST) (43), making the prediction of the function of NifZ impossible. Through this study, we were able to shed some light on this issue by proposing a chaperone-like function of NifZ in the formation of the second P cluster of the MoFe protein, although a detailed mechanism of the process (such as whether NifZ works alone or in combination with other components or how condensation of the two [4Fe-4S]-like clusters that ultimately leads to the for-  After the formation of one ␣␤ dimeric catalytic unit of the MoFe protein that contains one P cluster and one FeMoco, the two [4Fe-4S]-like clusters in the other ␣␤ dimer are forced into a conformation that does not allow proper cluster condensation that eventually leads to the formation of a P cluster. NifZ might be involved in this stage of the assembly process and acts in a chaperone-like fashion to facilitate a conformational rearrangement that is required for the condensation of these clusters into a P cluster and the subsequent insertion of FeMoco into the vacant FeMoco-binding site. mation of the P cluster takes place) awaits further investigation. Purification and characterization of the nifZ gene product are currently under way and will help to define the function of NifZ and to further refine our model of stepwise MoFe protein assembly.