INTRODUCTION

The reduction of atmospheric N₂ to ammonium by biological systems is catalyzed by the nitrogenase enzymes. The aerobe Azotobacter vinelandii harbors three genetically distinct nitrogenase enzymes that are regulated by the metal content of the growth medium, among other factors (1-3). Nitrogenases 1, 2, and 3 are encoded by the nif, vnf, and anf genes, respectively. nif-encoded nitrogenase 1 is a molybdenum-containing enzyme that is expressed in the presence of molybdenum. Expression of the vnf-encoded nitrogenase 2, a vanadium-containing enzyme, requires medium that is depleted in molybdenum and that contains vanadium. The anf-encoded nitrogenase 3 is expressed in medium deficient in both metals. All three nitrogenases are two-component metalloenzymes comprised of dinitrogenase and dinitrogenase reductase (1, 2). Dinitrogenase contains the active site metal center of the enzyme, and dinitrogenase reductase functions as the obligate electron donor to dinitrogenase during enzyme turnover in a MgATP- and reductant-dependent process (4, 5).

The active site of dinitrogenase 1, the iron-molybdenum cofactor (FeMo-co),¹ is composed of molybdenum, iron, sulfur, and homocitrate ((R)-2-hydroxyl-1,2,4-butanetricarboxylic acid) in a 1:7:9:1 ratio (6-9). The biosynthesis of FeMo-co involves the products of at least six nif genes, including nifQ, nifB, nifV, nifE, nifN, and nifH (9-12). The nifQ gene product might be involved in the formation of a molybdenum-sulfur precursor to FeMo-co (13), and the nifV gene product encodes homocitrate synthase (9, 10).² The product of NIFB, termed NifB-co, is an iron- and sulfur-containing precursor to FeMo-co (14, 15). Based on the amino acid sequence identity of NIFEN to NIFDK (the structural polypeptides of dinitrogenase 1), and the fact that NIFEN has been shown to bind NifB-co (15-17), NIFEN has been proposed to be a scaffold for FeMo-co assembly; however, the precise function of NIFEN in FeMo-co biosynthesis is unknown, as is the function of NIFH (dinitrogenase reductase 1).

In addition to being necessary for the biosynthesis of FeMo-co, the gene products of both nifV and nifB are required for the biosynthesis of the iron-vanadium cofactor (FeV-co) of dinitrogenase 2 and the putative iron-only cofactor (FeFe-co) of dinitrogenase 3 (18-21). Thus, homocitrate is presumed to be present as a component of FeV-co and FeFe-co, and NifB-co is believed to serve as the iron and sulfur donor to all three cofactors. Homologs of nifE and nifN have been identified in the vnf but not in the anf system (22); homologs of nifH exist in both molybdenum-independent systems (23, 24). Additional gene products required for FeV-co and FeFe-co biosynthesis have not been identified.

An in vitro system for the synthesis of FeMo-co that requires at least molybdate, homocitrate, an ATP-regenerating mixture, a source of reductant, NifB-co, NIFEN, and NIFH has been described (12, 14, 25, 26). The in vitro FeMo-co synthesis system utilizes molybdenum with high specificity as addition of 100-fold excess tungstate (a competitive inhibitor of the molybdenum transport system in Klebsiella pneumoniae) or vanadate do not significantly inhibit FeMo-co synthesis (12). The replacement of molybdenum with vanadium or iron in the cofactor during in vitro synthesis has not been achieved. The preferential incorporation of molybdenum into FeMo-co suggests that a component(s) involved in FeMo-co biosynthesis might exclusively select for molybdenum. The presence of a dinitrogenase reductase associated with each nitrogen fixation system makes that protein a likely candidate for specifying the heterometal incorporated into the respective cofactors of the nitrogenase enzymes.

NIFH has multiple roles in the nitrogenase 1 enzyme system. In addition to MgATP-dependent electron transfer to dinitrogenase during substrate reduction, NIFH is required for the biosynthesis of FeMo-co (10, 11) and for the maturation of apodinitrogenase 1 (a catalytically inactive form of dinitrogenase 1 that lacks FeMo-co) to its FeMo-co-activatable form (27, 28). In the latter process, NIFH is required for the association of the protein (a non-nif-encoded protein) (28) with ₂₂ apodinitrogenase 1 to form the FeMo-co-activatable ₂₂₂ hexamer (27). Some altered forms of NIFH that are unable to function as a reductant for nitrogenase-dependent substrate reduction are fully functional in FeMo-co biosynthesis and in the maturation of apodinitrogenase 1 (29-31), indicating that the characteristics of NIFH that enable it to function in nitrogenase turnover are not necessary for its role in the formation of active dinitrogenase.

In vivo studies by Joerger et al. (23) and Gollan et al. (32) suggest that NIFH supports FeV-co synthesis and that ANFH (the anfh gene product) supports FeMo-co synthesis. We utilized the in vitro FeMo-co synthesis assay system to definitively determine whether VNFH would function in FeMo-co biosynthesis; the ability of VNFH to replace NIFH in the formation of the FeMo-co-activatable ₂₂₂ form of apodinitrogenase 1 was also examined. Studies on the specificity of the incorporation of molybdenum into FeMo-co are discussed.

EXPERIMENTAL PROCEDURES

DEAE-cellulose was a Whatman DE52 product. Sephacryl S-100 and the Mono Q anion exchange column were from Pharmacia Biotech Inc. The fast protein liquid chromatography instrument was from LKB. Sodium dithionite (DTH) was purchased from Fluka Chemicals. Sodium metavanadate (NaVO₃, 99.995% purity), Tris base, and glycine were Fisher products. Acrylamide/bisacrylamide solution was obtained from Bio-Rad. All reagents used for A. vinelandii growth medium were of analytical grade or higher purity. Tetrathiomolybdate ((NH₄)₂MoS₄) was a gift from D. Coucouvanis, and [K₂(H₂O)₅][(VO₂)₂(R,S-homocitrate)₂]·H₂O was a gift from W. Armstrong (33). All other chemicals were from Sigma.

A. vinelandii strains DJ1030 (nifHnifB) (28), CA12 (nifHDK) (34), UW45 (nifB^[minus0]) (35), and CA117.30 (nifDKB) (36) have been described. All vessels used in preparing media and for cell culture were rinsed thoroughly in 4 N HCl and then in deionized water. Cultures (15 liters) of strain DJ1030 were grown in 20-liter polycarbonate carboys with vigorous aeration at 30 °C on Burk's medium that lacked sodium molybdate and contained 10 µM NaVO₃ (for derepression of the vnf system) and 40 µg of nitrogen/ml as ammonium acetate. The cultures were monitored for depletion of ammonium, following which they were derepressed for 4.5 h. The cells were concentrated using a Pellicon cassette system equipped with a filtration membrane (0.45 µm, Millipore Corp., Bedford, MA) and were centrifuged. The cell pellets were frozen in liquid N₂ and stored at 80 °C. Strain DJ1030 was grown on Burk's medium containing 1 mM sodium molybdate in place of NaVO₃ for derepression of the nif system. Strain UW45 was grown and derepressed on tungsten-containing medium (molybdenum free) as described previously (12). Strain CA117.30 was grown in 250-ml cultures on Burk's medium containing 10 µM NaVO₃; cells were concentrated by centrifugation, and derepression was initiated (for 4 h) by suspending the cell pellets in nitrogen-free Burk's medium. Cells were harvested by centrifugation and frozen as described above. Cell-free extracts were prepared by the osmotic shock method (6).

All buffers were sparged with purified N₂ (and degassed on a gassing manifold where appropriate) for 10-30 min, and DTH was added to a final concentration of 1.7 mM. All buffers used in column chromatography contained 0.2 mM phenylmethylsulfonyl fluoride and 0.5 µg/ml leupeptin. Buffers used in fast protein liquid chromatography were filtered through a 0.45-µm filter. Tris-HCl was at pH 7.4 unless stated otherwise.

All column chromatography steps except for fast protein liquid chromatography were performed at 4 °C. VNFH was purified from extract of strain DJ1030 (nifHnifB, vnf-derepressed) with modifications to the method described by Hales et al. (37). One hundred fifty ml of cell-free extract (from 50 g of cell paste) were applied to a 2.5 × 17-cm DEAE-cellulose column that had been equilibrated in buffer containing 0.1 M NaCl in 0.025 M Tris-HCl, pH 7.4. Following application of the extract, the column was washed with 2 bed volumes of buffer containing 0.125 M NaCl in 0.025 M Tris-HCl; VNFH was eluted using 0.22 M NaCl in 0.025 M Tris-HCl. The DEAE-cellulose fraction was concentrated by ultrafiltration using a XM100-A membrane, and the retentate (4 ml) was applied to a 2.5 × 78-cm Sephacryl S-100 column that had been equilibrated with 0.05 M NaCl in 0.025 M Tris-HCl. The column was developed with the same buffer, and the VNFH-containing fractions that exhibited the highest activity were concentrated by ultrafiltration (described above) and purified further on a Mono Q anion exchange column used in conjunction with a fast protein liquid chromatography system. Two ml (6.8 mg of protein) of the VNFH-containing retentate from the ultrafiltration cell was applied onto the Mono Q column that had been equilibrated with 0.15 M NaCl in 0.025 M Tris-HCl. The column was washed with 1 bed volume of the equilibration buffer, following which VNFH was eluted using a 20-ml increasing linear gradient from 0.15 to 0.4 M NaCl (in 0.025 M Tris-HCl, pH 7.4). VNFH eluted with 0.32 M NaCl in 0.025 M Tris-HCl. Active fractions were stored in 9-ml, serum-stoppered vials at 80 °C. The ability of VNFH to transfer electrons to dinitrogenase 1 was tested using the acetylene reduction assay for nitrogenase activity, and the results were consistent with the published results. VNFH was equally effective as NIFH in transferring electrons to dinitrogenase 1, consistent with the results of Chisnell et al. (34).

FeMo-co was prepared in N-methylformamide as described previously (6). The reactions were performed in 9-ml, serum-stoppered vials that were repeatedly evacuated, flushed with argon, and rinsed with 0.3 ml of 0.025 M Tris-HCl containing 1.7 mM DTH. The following components were added to the vials in the order indicated: 100 µl of 0.025 M Tris-HCl; 200 µl of an ATP-regenerating mixture (containing 3.6 mM ATP, 6.3 mM MgCl₂, 51 mM phosphocreatine, 20 units/ml creatine phosphokinase, and 6.3 mM DTH); 200 µl (3.8 mg protein) of extract of strain DJ1030 (nifHnifB, nif-derepressed) as a source of ₂₂ apodinitrogenase 1 and the  protein; and 10-50 µl (0.1 mg of protein) of the appropriate dinitrogenase reductase. The vials were incubated for 10 min at room temperature to allow the formation of ₂₂₂ apodinitrogenase 1. One hundred µl of anoxic 50% glycerol were added to the reactions to be analyzed by native PAGE, and these vials were placed on ice. Ten µl of a solution containing an excess of FeMoco were added to the remaining vials, which were incubated for 10 min at room temperature during which ₂₂₂ apodinitrogenase 1 was activated by FeMo-co to form dinitrogenase 1. Fifty nmol of (NH₄)₂MoS₄ (prepared in N-methylformamide containing 1.7 mM DTH) were added to the vials to prevent further FeMo-co insertion into apodinitrogenase 1. Activity of the newly reconstituted dinitrogenase 1 was monitored by the C₂H₂ reduction assay for nitrogenase (12). (NH₄)₂MoS₄ was excluded in certain control reactions, and 0.1 mg of the appropriate dinitrogenase reductase (that used in the insertion phase of the assay) was added in place of 0.1 mg of NIFH normally added during the C₂H₂ reduction phase of the assay (12).

Nine-ml serum vials were repeatedly evacuated, flushed with argon, and rinsed with buffer containing 1.7 mM DTH. Components were added to the vials in the following order: 100 µl of 0.025 M Tris-HCl, 10 µl of 1 mM Na₂MoO₄, 20 µl of 5 mM homocitrate (that had been treated with base to cleave the lactone, pH 8.0), and 200 µl of the ATP-regenerating mixture (defined above). The vials were incubated at room temperature for 10-20 min. Two hundred µl of extract (~3.8 mg protein) of either DJ1030 (nifHnifB, nif-derepressed) or CA12 (nifHDK, nif-derepressed), 25 µl of a solution containing NifB-co, and 10-50 µl (0.1 mg protein) of the appropriate dinitrogenase reductase were added to the vials. The vials were incubated at 30 °C for 30-90 min. Following this incubation, 100 µl of anoxic 50% glycerol were added to the reactions to be analyzed by anoxic native PAGE, and these vials were placed on ice. Five nmol of (NH₄)₂MoS₄ (prepared as described above) were added to the remaining vials to prevent further FeMo-co synthesis during the subsequent C₂H₂ reduction phase of the assay. The activity of the newly formed dinitrogenase 1 was monitored by the C₂H₂ reduction assay. (NH₄)₂MoS₄ was excluded from certain reactions to which 0.1 mg of the appropriate dinitrogenase reductase (that used in the synthesis phase of the assay) was added in place of 0.1 mg of NIFH normally added during the C₂H₂ reduction phase of the assay.

Proteins were resolved on anoxic native gels with a 7-14% acrylamide (37.1% acrylamide, 1% bisacrylamide) and 0-20% sucrose gradient. The electrophoresis buffer was N₂-sparged, 65 mM Tris-glycine (pH 8.5) containing 1.7 mM DTH. Gels were pre-electrophoresed for at least 60 min at 120 V for initial reduction, and proteins were electrophoresed for 1920 V-h (at 120 V) at 4 °C. One hundred µl of the reaction mixtures were applied onto the gel.

Polyclonal antibodies to NIFH and the  protein were raised in rabbits (the anti- protein antibodies were prepared and made available by Drs. Mary Homer and Gary Roberts). Immunoblotting and developing procedures have been described (38). The native gels were equilibrated in transfer buffer for at least 15 min prior to blotting.

Protein concentrations of cell-free extracts and purified proteins were measured using the bicinchoninic acid method (39).

RESULTS AND DISCUSSION

Ability of VNFH to Support the Maturation of Apodinitrogenase 1 to the FeMo-co-activatable ₂₂₂ Form

The association of the  protein with ₂₂ apodinitrogenase 1 to form the ₂₂₂ FeMo-co-activable species requires the presence of NIFH and nucleotide (27, 28), as diagrammed in Reaction 1. 

22        222        22+2

Apodinitrogenase 1

   

Apodinitrogenase 1

     

Dinitrogenase 1

(catalytically active)

The FeMo-co insertion assay and anoxic native PAGE were employed to test whether VNFH might replace NIFH in the maturation of apodinitrogenase 1. The results in Table I show that treatment of extract containing ₂₂ apodinitrogenase 1 and the  protein with equivalent levels of purified NIFH or VNFH resulted in similar levels of activity in the FeMo-co insertion assay, indicating that VNFH is as effective as NIFH in the conversion of ₂₂ apodinitrogenase 1 to the ₂₂₂ form. Nucleotide is necessary for the VNFH-dependent maturation process as is maturation supported by NIFH (27). Control reactions in which (NH₄)₂MoS₄ was not added to quench further FeMo-co insertion and which contained VNFH in both insertion and C₂H₂ reduction phases of the assay exhibited similar levels of activity as reactions to which NIFH was added (following (NH₄)₂MoS₄ addition) during the C₂H₂ reduction phase. Thus, the activities reported in Table I for reactions that contained VNFH in the insertion phase alone were not a result of NIFH functioning to attach the  protein to ₂₂ apodinitrogenase 1 during the C₂H₂ reduction phase of the assay. NIFH from another organism (Rhodospirillum rubrum) also supported activity in the FeMo-co insertion assay (Table I).

Table I. Ability of VNFH to function in the FeMo-co insertion assay and in in vitro FeMo-co synthesis Dinitrogenase reductasea FeMo-co insertionb FeMo-co synthesisc  MgATP +MgATP C2H4 formed/min/assay None 0.3 0.4 0.03 NIFH 0.5 14.6 29.8 VNFH 0.4 14.5 8.2 NIFH (R. rubrum)d 0.4 13.7 NDe a Assays contained 0.1 mg of the appropriate dinitrogenase reductase protein and 3.8 mg of extract of strain DJ1030 (nifHnifB, nif-derepressed) as a source of 22 apodinitrogenase and  protein. b FeMo-co insertion assays were performed as described under "Experimental Procedures"; activities are expressed as nanomoles of C2H4 formed/min/assay. c FeMo-co synthesis assays were performed as described under "Experimental Procedures"; activities are expressed as nanomoles of C2H4 formed/min/assay. d NIFH was purified from R. rubrum as described in Ludden and Burris (42). e Not determined.

To confirm the results of the FeMo-co insertion assays, we employed anoxic, native PAGE to monitor the association of the  protein with ₂₂ apodinitrogenase 1 in extracts of strain DJ1030 (nifHnifB, nif-derepressed) in the presence of nucleotide and the different dinitrogenase reductase proteins. Fig. 1, an immunoblot of an anoxic, native gel (developed with antibody to the  protein), illustrates that VNFH functions in the association of the  protein with ₂₂ apodinitrogenase 1 (Fig. 1, lane 3). These results are consistent with the activities observed in the FeMo-co insertion assays testing the different dinitrogenase reductase proteins (Table II).

Immunoblot (developed with antibody to the  protein) of an anoxic native gel illustrating the association of the  protein with ₂₂ apodinitrogenase 1.

Table II. Specificity of the incorporation of molybdenum into FeMo-co Extract of straina Metal addedb Cofactor synthesisc nmol C2H4 formed/min/assay UW45 (nifB, tungsten-grown) None 0.1 As above Molybdenum 18.3 As above Vanadium 0.06 As above Iron 0.08 CA117.30 (nifDKB vnf-derepressed) Molybdenum 0.02 As above Vanadium 0.02 a Two hundred µl (3.8 mg of protein) of extract of strain UW45 (tungsten-grown) were used in the FeMo-co synthesis assays as a source of all nif-encoded proteins. Two hundred µl (3.6 mg of protein) of extract of strain CA117.30 (vnf-derepressed) was used as a source of vnf-encoded proteins for the synthesis of FeV-co. b Molybdenum was included the assays in the form of Na2MoO4; vanadium was added to the assays as NaVO3, V2O5, VCl3, VOPO4, or [K2(H2O)5][(VO2)2(R,S-homocitrate)2]·H2O. Iron was included in the assays as FeNO3 or FeCl3. c Activities are expressed as nanomoles of C2H4 formed/min/assay.

The high degree of amino acid sequence identity between NIFH and VNFH (91%) (1) is consistent with the effectiveness of VNFH in both substrate reduction (when complemented with dinitrogenase 1) and in the maturation of apodinitrogenase 1. The domain(s) of NIFH required for both the above functions are quite likely highly conserved in VNFH. At present, the role(s) of the dinitrogenase reductase protein in the maturation of apodinitrogenase 1 remains under investigation.

VNFH was tested in the in vitro FeMo-co synthesis assay in place of NIFH (Table I). VNFH typically exhibited 25-30% of the FeMo-co synthesis activity (in our fixed time assay) observed with an equivalent level of NIFH, despite exhibiting similar levels of activity in the C₂H₂ reduction assay. Addition of increasing levels of VNFH and increasing the time allowed for in vitro FeMo-co synthesis did not result in a linear increase in activity (data not shown). The limiting step(s) in the assay is not the maturation of apodinitrogenase 1, because VNFH functions as effectively as NIFH in the maturation process (discussed above). The reasons for the lower level of FeMo-co synthesis observed with VNFH are not known. It is possible that VNFH is unable or slow to dissociate from a nif protein(s) with which it interacts during the course of FeMo-co synthesis, thus limiting further turnover of the protein(s) involved.

Homer et al. (28) demonstrated that the  protein dimer (present in extracts of A. vinelandii strains unable to synthesize FeMo-co) monomerized upon associating with FeMo-co, and thus it was possible to employ the monomerization of the  protein (detected by anoxic native PAGE) as an alternate assay for the completion of FeMo-co synthesis. Thus, FeMo-co synthesized in vitro in reaction mixtures containing an extract of strain CA12 (nifHDK, nif-derepressed) would accumulate on the  protein (resulting in the monomerization of the  protein dimer) due to the absence of apodinitrogenase 1 in extracts of this strain. Fig. 2 is an immunoblot (developed with antibody to the  protein) of an anoxic native gel that demonstrates the results of this study. When dinitrogenase reductase is excluded from the in vitro FeMo-co synthesis reaction, the  dimer and a slow migrating species of  that is uncharacterized (indicated by X on Fig. 2) are observed (Fig. 2, lane 1); the dimeric form of the  protein is observed in extracts of strains that are impaired in FeMo-co biosynthesis (33). That both NIFH and VNFH support FeMo-co biosynthesis is illustrated by the monomerization of the  protein observed as the faster migrating  protein-FeMo-co form in reactions that included NIFH or VNFH (Fig. 2, lanes 2 and 3).

Immunoblot (developed with antibody to the  protein) of an anoxic native gel demonstrating the monomerization of the protein upon FeMo-co binding.

Does dinitrogenase reductase specify the heterometal contained in the nitrogenase cofactors? Two lines of evidence suggest that the dinitrogenase reductases do not specify or select against the heterometal that is incorporated into the cofactors of the nitrogenase enzymes: 1) the ability of VNFH to function in in vitro FeMo-co synthesis (albeit less effectively than NIFH), and 2) the observation by Joerger et al. (23) that NIFH supported vanadium-dependent diazotrophic growth of an A. vinelandii strain containing a deletion in the vnfH gene, indicating that, in vivo, NIFH functions in FeV-co biosynthesis. Gollan et al. (32) demonstrated the in vivo synthesis and incorporation of FeMo-co into the dinitrogenase 3 polypeptides of a Rhodobacter capsulatus strain containing deletions in the nifHDK genes; the synthesis of FeMo-co in the absence of a nifH gene suggests that ANFH most likely replaced NIFH in the synthesis of FeMo-co. Our results demonstrating the ability of VNFH to function in the in vitro biosynthesis of FeMo-co suggest that the dinitrogenase reductase protein quite likely does not select against the incorporation of molybdenum into FeV-co and FeFe-co.

Cofactor structures of the three nitrogenases are proposed to be essentially similar with vanadium and iron atoms replacing the molybdenum atom in FeV-co and FeFe-co, respectively (2, 21, 40). The requirement of the nifB and nifV gene products for the biosynthesis of all three cofactors suggests that certain steps in the biosynthesis of FeMo-co are shared in the biosynthetic pathways of all three cofactors. Although FeV-co is largely uncharacterized, extended x-ray absorption fine structure studies on dinitrogenase 2 indicate that FeV-co is similar in structure to FeMo-co with the octahedral vanadium atom surrounded by 3 oxygen atoms and 3 sulfur atoms as is the molybdenum atom in FeMo-co (41). Other similarities between FeMo-co and FeV-co include the ability to extract FeV-co into N-methylformamide (20) and its probable ligation to the dinitrogenase 2 polypeptide via the conserved cysteine and histidine residues (analogous to Cys-275 and His-442 of NIFD) that ligate FeMo-co to dinitrogenase 1 (8, 23).

To determine whether the FeMo-co synthesis system would utilize vanadium and iron in the synthesis of FeV-co and FeFe-co, respectively, we tested various vanadium- and iron-containing compounds in place of molybdenum in the in vitro FeMo-co synthesis assay. Extract of A. vinelandii strain UW45 (nifB, tungsten-grown) was used as a source of all the nif-encoded proteins necessary for the synthesis of FeMo-co. Active dinitrogenase 1 was formed only when molybdenum (in the form of Na₂MoO₄) was included in the in vitro reactions (Table II). Molybdenum added to in vitro FeMo-co synthesis reactions in the form of (NH₄)₂MoO₂S₂, K₂MoO₃S, and MoS₂ also supported in vitro FeMo-co synthesis (data not shown). Vanadium added in the form of NaVO₃, V₂O₅, VCl₃,VOPO₄, or [K₂(H₂O)₅][(VO₂)₂(R,S-homocitrate)₂]·H₂O did not produce active dinitrogenase 1. Similar results were obtained when iron (in the form of FeCl₃ and Fe(II)NO₃) was included in the assay. Several possibilities might account for these results. The FeMo-co synthesis machinery might indeed discriminate against vanadium and iron; however, in vivo studies demonstrating the ability of NIFEN and NIFH to support vanadium-dependent diazotrophy suggest that certain nif proteins required for FeMo-co biosynthesis do function in FeV-co biosynthesis in vivo (22, 23). Vanadium and iron might not be in their correct oxidation states or precursor forms necessary for incorporation into the cofactor under the in vitro assay conditions.

We employed cell-free extracts of strain CA117.30 (nifDKB) that was derepressed on vanadium to determine whether FeV-co could be synthesized under conditions similar to those used to synthesize FeMo-co in vitro. When extract of CA117.30 (nifDKB, vnf-derepressed) was used as a source of vnf-encoded proteins in in vitro reactions containing vanadium (in the form of NaVO₃, V₂O₅, VCl₃, VOPO₄, or K₂(H₂O)₅][(VO₂)₂(R,S-homocitrate)₂]·H₂O), homocitrate, ATP (in the form of an ATP-regenerating mixture), and NifB-co, formation of active dinitrogenase 2 was not observed (Table II). Varying the nucleotides included in the reactions, the pH of the reaction mixture, and addition of partially purified apodinitrogenase 1 and NIFEN (in case of limiting levels of apodinitrogenase 2 and VNFEN in the vnf-derepressed extracts) to certain reactions also produced negative results. Clearly, the in vitro conditions under which FeMo-co is synthesized are inadequate for the synthesis of FeV-co. As discussed above, the conversion of vanadium to the form required for its incorporation into FeV-co might not occur in vitro; alternatively, intermediates in the FeV-co biosynthetic pathway might be unstable under our cell-breakage and assay conditions. These observations suggest that steps and precursors unique to the synthesis of FeV-co quite likely exist. The identification of additional vnf genes and the characterization of phenotypes of strains carrying lesions in vnf genes might enable the elucidation of steps involved in the biosynthesis of FeV-co.