Cloning and Mutational Analysis of the (cid:1) Gene from Azotobacter vinelandii Defines a New Family of Proteins Capable of Metallocluster Binding and Protein Stabilization*

Dinitrogenase is a heterotetrameric ( (cid:2) 2 (cid:3) 2 ) enzyme that catalyzes the reduction of dinitrogen to ammonium and contains the iron-molybdenum cofactor (FeMo-co) at its active site. Certain Azotobacter vinelandii mutant strains unable to synthesize FeMo-co accumulate an apo form of dinitrogenase (lacking FeMo-co), with a subunit composition (cid:2) 2 (cid:3) 2 (cid:1) 2 , which can be activated in vitro by the addition of FeMo-co. The (cid:1) protein is able to bind FeMo-co or apodinitrogenase independently, leading to the suggestion that it facilitates FeMo-co insertion into the apoenzyme. In this work, the non- nif gene encoding the (cid:1) subunit ( nafY ) has been cloned, sequenced, and found to encode a NifY-like protein. This finding, together with a wealth of knowledge on the biochemistry of proteins involved in FeMo-co and FeV-co biosyntheses, allows us to define a new family of iron and molybdenum (or vanadium) cluster-binding proteins that includes NifY, NifX, VnfX, and now (cid:1) . In vitro FeMo-co insertion experiments presented in this work demonstrate that

Nitrogenase catalyzes the reduction of nitrogen gas to ammonium, in an ATP-and reductant-dependent reaction. It is one of the best characterized metalloenzymes and is an excellent model for elucidating metalloprotein assembly. Nitrogenase is composed of two oxygen-labile metalloproteins: dinitrogenase and dinitrogenase reductase (1,2). Dinitrogenase (also termed component I or molybdenum-iron protein) is a 240-kDa ␣ 2 ␤ 2 tetramer of the nifD and nifK gene products (3). Each ␣␤ nitrogenase dimer contains an iron-molybdenum cofactor (FeMo-co) 1 and a P cluster (3,4). Dinitrogenase reductase (also termed component II or iron protein) is a 60-kDa ␣ 2 dimer of the nifH gene product which contains a single 4Fe-4S center coordinated between the two subunits (5). NifH has at least three roles in the nitrogenase enzyme system (6): first, it serves as electron donor to nitrogenase; second, it participates in the biosynthesis of FeMo-co; and third, it is required for maturation of apodinitrogenase to a FeMo-co-activable form.
The genes that encode dinitrogenase (nifD and nifK) are not required for FeMo-co biosynthesis, suggesting that FeMo-co is assembled elsewhere in the cell and is then inserted into apodinitrogenase (7). It is known that the products of at least seven nitrogen fixation (nif) genes, nifB, nifE, nifH, nifN, nifQ, nifV, and nifX, are involved in the biosynthesis of FeMo-co (8).
Azotobacter vinelandii or Klebsiella pneumoniae strains with mutations in nifB, nifN, or nifE produce a FeMo-co-deficient hexameric (␣ 2 ␤ 2 ␥ 2 ) apodinitrogenase that can be activated in vitro by the simple addition of purified FeMo-co (9,10). On the other hand, apodinitrogenase from ⌬nifH mutants has a tetrameric composition (␣ 2 ␤ 2 ) and requires some type of NifH-and MgATP-dependent maturation that, in turn, promotes the association of the ␥ subunit and leads to the form that is competent for FeMo-co activation (11,12).
In vitro studies on crude extracts of an A. vinelandii nifB mutant demonstrated that the ␥ protein specifically binds to free FeMo-co and to apodinitrogenase, consistent with a role in FeMo-co insertion (13). However Christiansen et al. (14) have reported that pure preparations of His-tagged apodinitrogenase from a nifB mutant strain lacked the ␥ subunit but still could be activated to 80% of the theoretical value by the simple addition of FeMo-co, which suggests a role for ␥ other than that of a FeMo-co insertase. As they pointed out, the solution to this controversy would require the inactivation of the ␥ gene.
In K. pneumoniae the third subunit in the hexameric apodinitrogenase is the product of the nifY gene (10,15). However, this is not the case in A. vinelandii, where the ␥-encoding gene is not in the nif gene clusters. Although A. vinelandii contains a nifY gene there is no mutant phenotype associated with its inactivation (16).
In this work, we have cloned and inactivated the A. vinelandii gene encoding the ␥ protein. In vivo and in vitro results presented here indicate that ␥ stabilizes apodinitrogenase in an open (FeMo-co activable) conformation; by contrast, our * This work was supported by Grant 35332 from the NIGMS, National Institutes of Health (to P. W. L.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM  results do not support a role for ␥ as an essential FeMo-co insertase.

EXPERIMENTAL PROCEDURES
Materials-Sodium dithionite was from Fluka. Leupeptin, phenylmethylsulfonyl fluoride, phosphocreatine, creatine phosphokinase, and ATP were from Sigma. Tris and glycine were from Fisher Scientific. Nitrocellulose and polyvinylidene difluoride membranes were from Millipore. Acrylamide/bisacrylamide and the equipment for SDS-PAGE were from Bio-Rad. Ammonium tetrathiomolybdate ((NH 4 ) 2 MoS 4 ) was a gift from D. Coucouvanis (University of Michigan, Ann Arbor).
Buffers-25 mM Tris⅐HCl, pH 7.5, was used throughout this work. All buffers for protein analysis were sparged with purified N 2 for 20 -30 min, and sodium dithionite was added to a final concentration of 1 mM. Buffers used for obtaining A. vinelandii cell-free extracts contained 0.5 g/ml leupeptin and 0.2 mM phenylmethylsulfonyl fluoride.
Strains and Growth Conditions-A. vinelandii strains DJ (wild type), DJ166 (⌬nifX), and DJ208 (⌬nifY) were obtained from D. R. Dean, Department of Biochemistry, Virginia Technical Institute, Blacksburg, Virginia. Strain UW45 (nifB) has been described previously (17). Growth in the presence of molybdate, nif derepression, and cell breakage have been described (18). For growth on plates, the medium was solidified with separately autoclaved 1.5% agar solution. 0.5 g/ml kanamycin, 0.25 g/ml streptomycin, 10 g/ml spectinomycin, and 5 g/ml rifampin were added as required. Escherichia coli DH5␣ was grown in Luria-Bertani medium at 37°C with shaking (200 rpm). For growth of E. coli on plates, medium solidified with 1.5% agar was used. Antibiotics were used at standard concentrations (19).
For growth rate determinations, strains were grown at 30°C on Burk's modified medium (designated as standard growth conditions) or at 37°C on Burk's modified medium not supplemented with molybdate (designated as stressing growth conditions). When a fixed nitrogen source was required, ammonium acetate was added to a final concentration of 29 mM. Growth was estimated from the absorbance of the cultures at 600 nm, using a Shimadzu UV1201V spectrophotometer. The growth rate constant corresponds to ln2/t d , where t d represents the doubling time.
Cloning of the ␥-Encoding Gene (nafY)-The N-terminal amino acid sequence for ␥ (VTPVNMSRETALRIALAARALPGTTVGQLL) was obtained from a preparation of partially purified ␣ 2 ␤ 2 ␥ 2 apodinitrogenase (see below). Degenerate oligonucleotides 5Ј-GTNACNCCNGTNAT-CATG-3Ј and 5Ј-CTGICCIACGGTGGTICCIGG-3Ј were designed, based on the N-terminal amino acid sequence, and were used as primers in a PCR to amplify an 81-bp DNA fragment from the chromosome of A. vinelandii. After sequencing the 81-bp fragment, oligonucleotides 5Ј-CGCCCTGGCTGCCAGGGCTTTG-3Ј (complementary to nucleotides 42-63 with respect to the translation start of the ␥ gene, termed nafY for nitrogenase accessory factor Y) and 5Ј-ATGCGCAGAGCGGTT-TCGCGAC-3Ј (complementary to nucleotides 41-20 with respect to the translation start of nafY) were used as primers in a reverse PCR. SalI-digested and religated chromosomic DNA from A. vinelandii was used as template for this reaction. A 1.5-kbp PCR product was obtained, ligated into pGEM-T, sequenced, and digested with BglII and SalI to generate a 1,236-bp DNA fragment that was finally ligated into pUK21 to render plasmid pRHB20. Plasmid pRHB20 contains sequences 5Ј of nafY from the A. vinelandii chromosome. A. vinelandii strain UW139 was generated by transforming DJ strain with plasmid pRHB20 and selecting for single recombinants on plates containing ammonium acetate-and kanamycin-supplemented Burk's medium. Genomic DNA from strain UW139 was digested with either BamHI or XhoI, religated, and used to transform E. coli DH5␣, rendering plasmids pRHB21 and pRHB24, respectively. Plasmid pRHB21 contains Ϸ14 kbp of A. vinelandii DNA sequences 5Ј of nafY, whereas plasmid pRHB24 contains 1.2 kbp of sequences 5Ј of nafY along with nafY and Ϸ16 kbp of 3Ј-sequences (see Fig. 1).
Plasmid Constructions and DNA Manipulations-Plasmid constructions, PCR, and transformation of E. coli were carried out by standard methods (19). A computer search for homologies was made using the BLAST algorithm (20). Isolation of DNA from A. vinelandii strains was carried out using the DNAeasy TM Tissue Kit (Qiagen).
RNA Isolation and Northern Analysis-For isolation of RNA, A. vinelandii strains grown in modified Burk's medium containing fixed nitrogen (29 mM ammonium acetate) were derepressed for nitrogenase as described (21). Isolation of RNA from A. vinelandii strains was performed using the RNeasy TM Midi Kit (Qiagen). For Northern analysis, 10-g portions of RNA samples were electrophoresed in a denaturing formaldehyde gel and transferred to a positively charged nylon membrane (Millipore). Prehybridization and hybridization were performed in the presence of 50% formamide at 42°C. A PCR-amplified DNA fragment, covering the entire nafY, was used as probe after labeling with Prime-It II Random Primer Labeling Kit (Stratagene) and [␣-32 P]dCTP.
Mutagenesis of A. vinelandii Genes-Procedures for A. vinelandii transformation (22) and gene replacement (23) have been described. Plasmid pRHB25 contains a 3.4-BglII DNA fragment from plasmid pRHB24 which includes a portion of rnfH, nafY, and additional 3Јsequences. Plasmids pRHB29a and pRHB29b were generated by substitution of a kanamycin resistance cassette for a 231-bp SalI fragment within nafY in plasmid pRHB25. The kanamycin resistance gene in the cassette, obtained from plasmid pUC4K, was in the same orientation as nafY in plasmid pRHB29a and in the opposite orientation from nafY in plasmid pRHB29b. UW146 was generated by transformation of strain UW45 with plasmid pRHB29b. Strains UW141, UW147, and UW154 were generated by transformation of strains DJ, DJ166, and DJ208 with plasmid pRHB29a, respectively, followed by selection of a Kan r phenotype. Plasmid pRHB28 was generated after removing the 231-bp SalI fragment from pRHB25 described above, to produce an in-frame deletion within nafY. Strains UW149, UW156, and UW158 were generated by transformation of strains UW141, UW154 and UW147 with plasmid pRHB28, respectively, and scored for the Kan s phenotype. UW166 was generated by transformation of UW156 with plasmid pRHB44a, which contains nifX disrupted by an insertion of a kanamycin resistance cassette at the internal Eco47III site to nifX. The kanamycin resistance gene in the cassette, obtained from plasmid pUC4K, was in the same orientation as nifX. Plasmid pRHB42 contains a 1.5-kbp SalI fragment from pRHB24 which includes a portion of rnfG along with rnfE, rnfH, and a portion of nafY. Plasmid pRHB43 was generated by insertion of the ⍀ interposon, from plasmid pHP45⍀, into the unique BglII site within rnfH in pRHB42. UW165 was generated by transformation of strain DJ with plasmid pRHB43. All generated mutations were checked by PCR to confirm that double recombination and segregation events occurred.
In Vitro Dinitrogenase and Dinitrogenase Reductase Activities-Dinitrogenase and dinitrogenase reductase activities in cell-free extracts were obtained after titration with an excess of the complementary component as described (24). The specific activity of each protein is defined as nmol of ethylene formed/min/mg of protein.
Activation of Apodinitrogenase with Purified FeMo-co (in Vitro FeMo-co Insertion Assay)-FeMo-co was prepared in N-methylformamide as described (25). 9-ml serum vials were evacuated and flushed repeatedly with purified argon and rinsed with anaerobic 25 mM Tris⅐HCl buffer, pH 7.5. The following were then added to the vials: 100 l of Tris⅐HCl buffer, 200 l of UW45 or UW146 cell-free extracts (Ϸ3 mg of protein), and 2 or 10 l of a solution containing FeMo-co (equivalent to 0.1 and 0.5 nmol of Mo, respectively). The mixtures were incubated for 30 min at 30°C after which 0.8 ml of ATP-regenerating mixture (containing 3.6 mM ATP, 6.3 mM MgCl 2 , 51 mM phosphocreatine, 20 units/ml creatine phosphokinase, and 6.3 mM sodium dithionite) and an excess of purified NifH (0.2 mg of protein) were added. Nitrogenase activity was then quantitated by acetylene reduction as described (26).
For the in vitro FeMo-co insertion time course experiments, 21-ml serum vials were evacuated and flushed repeatedly with purified argon and rinsed with anaerobic Tris⅐HCl buffer. The following were then added to the vials: 0.7 ml of Tris⅐HCl buffer, 1.4 ml of UW45 or UW146 cell-free extracts (Ϸ21 mg of protein), and 70 l of a solution containing FeMo-co (equivalent to 3.5 nmol of molybdenum). The mixtures were incubated at 30°C, and 0.2-ml aliquots were removed at different times and injected into anaerobic 9-ml serum vials containing 40 nmol of (NH 4 ) 2 MoS 4 to prevent further insertion of FeMo-co into apodinitrogenase. The vials were incubated for at least 10 min at room temperature, and 0.8 ml of ATP-regenerating mixture and an excess of purified NifH (0.2 mg of protein) were added. Nitrogenase activity was then quantitated by acetylene reduction as described (26).
SDS-PAGE and Immunoblot Analysis-The procedure for SDS-PAGE has been described (27). Immunoblot analysis was performed as described by Brandner et al. (28).
Preparation of ␥ Protein for N-terminal Sequencing-A preparation of partially purified hexameric apodinitrogenase (␣ 2 ␤ 2 ␥ 2 ) from strain UW45, containing about 12 g of protein, was subjected to SDS-PAGE and transferred to a polyvinylidene difluoride membrane. The ␥ protein band was excised from the membrane and used for N-terminal sequence at the Protein/Peptide Micro Analytical Laboratory of the California Institute of Technology.
Protein Assays-Protein concentrations were determined by the bicinchoninic acid method using bovine serum albumin as standard (29).

RESULTS AND DISCUSSION
Cloning of the Gene Encoding the ␥ Protein from A. vinelandii-A combination of PCR and reverse PCR techniques was used to isolate the gene encoding ␥ from A. vinelandii genomic DNA (for details see "Experimental Procedures"). The cloning procedure resulted in the recovery of plasmids pRHB21 and pRHB24. pRHB21 contains Ϸ14 kbp of A. vinelandii DNA sequence 5Ј of the ␥-encoding gene (nafY), whereas pRHB24 contains 1.2 kbp of sequence 5Ј of nafY along with nafY and Ϸ16 kbp of 3Ј-sequences (Fig. 1). Sequencing of the 1.2-kbp SalI-BglII fragment located 5Ј of nafY revealed the presence of three ORFs, designated ORF1, ORF2, and ORF3, which would encode polypeptides showing homology to RnfG, RnfE, and RnfH polypeptides from Rhodobacter capsulatus, respectively (30). In R. capsulatus, the Rnf polypeptides are required in vivo for nitrogen fixation and are proposed to constitute a membrane complex involved in electron transport to nitrogenase (30,31). Thus, rnf genes and nafY are clustered in the A. vinelandii chromosome (Fig. 1). No other ORFs were found in the 500 bp 3Ј of nafY in plasmid pRHB24, strongly suggesting that nafY is the last gene of the cluster.
nafY Encodes a NifY-like Protein- Fig. 2 compares the amino acid sequences of ␥ with some ␥ homologs found in protein data bases. It is clear that nafY encodes a protein that is similar to NifY, NifX, VnfX, and the C-terminal half of NifB from A. vinelandii and other bacterial sources. The actual role of NifY in the A. vinelandii nif system is not known. However, K. pneumoniae NifY is found instead of ␥ as the third subunit in the hexameric apodinitrogenase (10,15). NifX and VnfX are proteins involved in the biosyntheses of FeMo-co and FeV-co, respectively (8,32). FeV-co is an iron-vanadium cofactor contained in the vnf-encoded dinitrogenase and functions analogously to FeMo-co of the molybdenum system (33). NifB is required for the three nitrogenase systems (molybdenum-, vanadium-, and iron-only-containing nitrogenases) (34), and its role may be the synthesis of NifB-co, an iron-sulfur cluster that serves as precursor for FeMo-co, FeV-co, and FeFe-co biosyntheses (35). Sequence identity between nifY and nifX gene products from A. vinelandii and K. pneumoniae have been noted before (16), but no functional relationships could be inferred at that time. When sequence identity between NifY/NifX and NifB proteins was noted, a role for NifY/NifX proteins in FeMo-co maturation or stabilization was suggested based on the known role of NifB in FeMo-co biosynthesis (36). Interestingly, the most similar protein in data bases is the recently reported product of a nifY-like gene in the bacterium Pseudomonas stutzeri (GenBank accession no. AJ297529). The physical organization of the rnf region in P. stutzeri (rnfCDGEH-nifY-like-ORF13-ORF12-nifH) somewhat resembles the A. vinelandii organization, where two separate rnfGEH-nafY (this work) and ORF13-ORF12-nifH regions have been found (GenBank accession no. AF014048). Although it is reasonable to believe that there might be a physical linkage between the two regions in the A. vinelandii chromosome, this is unlikely because we have not been able to complement the mutation in A. vinelandii strain DJ54 (⌬nifH) by transformation with plasmids pRHB21 or pRHB24.
Consistent with the sequence similarity among ␥, NifY, NifX, and VnfX polypeptides, in vitro studies provide supporting evidence that they are functionally related because they are able to interact with some intermediates of the FeMo-co or FeV-co biosynthetic pathways. First, the ␥ protein binds FeMo-co (13). Second, NifX and VnfX from A. vinelandii are also able to bind FeMo-co (37), 2 as well as NifB-co (32,37). Third, VnfX is also able to bind a FeV-co precursor (vanadiumcontaining iron-sulfur cluster) lacking homocitrate (32). Taken together, these lines of evidence indicate that ␥, NifY, NifX, and VnfX constitute a family of iron and molybdenum (or vanadium) cluster-binding proteins.
nafY and rnfH Are Transcribed Independently-It has been reported that the rnf gene cluster from R. capsulatus is subject to nif regulation and that their products are required in vivo for nitrogen fixation (30). In contrast, nothing is known about regulation of the expression of the rnf gene cluster in A. vinelandii, but it is known that ␥ (NafY) is not the product of a nif gene, nor is it nif coregulated (13). Because rnfH and nafY are clustered in the A. vinelandii chromosome, having an intergenic region of only 24 bp, we wondered whether they are cotranscribed. As a preliminary approach to address this question, the ⍀ interposon was inserted into the BglII site located within rnfH to generate a polar mutation (Fig. 1). The ⍀ interposon carries transcriptional terminators and has been shown to impair gene expression from promoters 5Ј of the site of ⍀ insertion in a wide range of Gram-negative bacteria (38). Crude extracts were prepared from nif-derepressed cells of wild-type and mutant strain UW165 (rnfH::⍀) and analyzed by immunoblot of an SDS-gel developed with antibody to ␥ (Fig. 3). The presence of ␥ in the extracts of strain UW165 suggests that a promoter 3Ј of the ⍀ interposon insertion site is driving nafY expression. However, the 80-bp region between the ⍀ insertion site and the nafY start codon lacks any obvious promoter sequences and is unable to promote ␥ accumulation when present in a plasmid in E. coli (data not shown). Given the substantial amount of ␥ protein that A. vinelandii cells are able to accumulate (9, 13), the lack of an obvious promoter sequence 5Ј of nafY is unexpected; nevertheless the result still suggests independent regulation of rnfH and nafY.
In addition, a Ϸ1.1-kb RNA species was identified by Northern analysis using a nafY probe (Fig. 4). The observed transcript is of sufficient length to encode the whole NafY and is present in both ammonium-grown and nif-derepressed cells (compare lanes 1 and 2), although it appears to be more abundant in nif-derepressed cells. Consistent with the immunoblot results, cells from strain UW165 (rnfH::⍀, lanes 3 and 4) also contain the 1.1-kb nafY transcript, indicating independent transcription from rnfH. Our Northern analysis does not, however, rule out the presence of additional longer transcripts that could cover part or the whole rnf-nafY cluster.
Mutational Analysis of the nafY Gene-The nafY gene was mutated by creation of a nonpolar in-frame deletion, termed ⌬nafY, and the mutation was transferred to the chromosome of A. vinelandii wild-type strain to generate strain UW149 (see "Experimental Procedures"). Immunoblot analysis showed the absence of ␥ in extracts of UW149 (Fig. 3). As presented in Table I, molybdate-dependent diazotrophic growth rate was minimally affected in strain UW149. Also, crude extracts of strain UW149 exhibit normal levels of in vitro dinitrogenase and dinitrogenase reductase activities (Table II). These results suggest that the role of ␥ is not essential for diazotrophic growth under standard laboratory conditions, perhaps because some other protein is performing the same function when ␥ is not present.
To test the possibility of functional redundancy because of the presence of NifY or NifX (which have some sequence and functional similarity as noted above), ⌬nafY mutation was transferred to the chromosome of strains DJ208 (⌬nifY) and DJ166 (⌬nifX), generating UW156 (⌬nifY⌬nafY) and UW158 (⌬nifX⌬nafY), respectively. UW166 (⌬nifY⌬nafYnifX::kan) was subsequently constructed by transformation of strain UW156 with plasmid pRHB44a (for details, see "Experimental Procedures"). As in the case of ⌬nafY mutant (see above), A. vinelandii ⌬nifX and ⌬nifY mutants are not impaired for diazotrophic growth under the normal diazotrophic growth conditions (16). A 30% decrease in the growth rates under standard diazotrophic growth conditions was observed in strains bearing ⌬nafY or ⌬nifX mutations (Table I). On the other hand, the diazotrophic growth rate was unaffected in the ⌬nifY mutant. Nitrogenase-derepressed extracts of these mutants were used to determine the effect of combining nafY, nifY, and nifX mutations on the levels of dinitrogenase activity. As shown in Table II, dinitrogenase activity levels are not altered significantly in any of the mutants when grown under standard nitrogen fixation conditions. In short, the data presented in Tables I and II show that when A. vinelandii is grown at 30°C with sufficient molybdenum, there is no additive effect of mutations in any two or all three of the members of the nafY/nifY/ nifX family. In all cases, a small reduction of diazotrophic growth rate is observed, and similar levels of dinitrogenase activity are found in crude extracts. It is possible that in the standard diazotrophic growth conditions used in the laboratory, in vivo FeMo-co synthesis and insertion could be sufficiently good for adequate growth to make the nafY/nifY/nifX function dispensable.
Following this logic, we attempted to force A. vinelandii cells to maintain a stable apodinitrogenase at a higher temperature and in an "open" conformation by reducing FeMo-co availability inside the cell. Thus, A. vinelandii strains carrying mutations in nafY, nifY, and nifX genes were tested for their abilities to grow under conditions of elevated temperature (37°C) and molybdenum stress. Molybdenum stress refers to cells growing in Burk's medium not supplemented with the standard 10 M Na 2 MoO 4 . A. vinelandii is known to be extremely efficient in scavenging traces of molybdenum from the culture medium (39), and the short-term lack of a molybdenum supplement should not be interpreted as a Mo-free medium. Under these conditions, the combination of mutations in nafY and nifX (UW158) and nafY, nifY, and nifX (UW166) had a profound effect on diazotrophic growth rates (5-fold lower than wild type), whereas UW149 (⌬nafY), DJ166 (⌬nifX), and UW156 (⌬nafY⌬nifY) strains exhibited decreases of Ϸ35% in their growth rates, and no effect was observed for DJ208 (⌬nifY) ( Table I). In contrast, all mutant strains exhibited wild-type growth rates when using ammonium as nitrogen source under the same culturing conditions at 37°C (data not shown).
The mutation in nafY also had an effect on the levels of extractable dinitrogenase activity when mutant strains were grown under stressing nitrogen fixation conditions (Table II). Dinitrogenase activity in the wild-type strain (DJ) was 60% of the value observed in standard conditions. However, it was severely diminished (2-5% of the values observed in standard conditions) in strains carrying mutations in nafY (UW149), nafY and nifX (UW158), and nafY, nifX, and nifY (UW166). This decrease in the activity of dinitrogenase in nafY mutant strains correlated with the accumulation of lower levels of NifDK polypeptides, as checked by immunoblot (data not shown). It is clear that the function of nafY is very important for dinitrogenase activity when cells are under stressing nitrogen fixation conditions. It is important to note that no ethane was detected during the acetylene reduction assays performed to determine nitrogenase activities under molybdenum stressing conditions. The coproduction of ethylene and ethane from acetylene by nitrogenase is characteristic of alternative molybdenum-independent nitrogenases and, therefore, the result indicates that only the molybdenum-dependent nitrogenase system is present. Table III shows that the addition of purified FeMo-co to cell FIG. 4. Identification of nafY transcripts from strains DJ (wild-type) and UW165 (rnfH::⍀). RNA isolated from A. vinelandii cells grown with ammonium (lanes 1 and 3) or derepressed for nitrogen fixation (lanes 2 and 4) was probed with a DNA fragment covering the entire nafY. The arrow points to a Ϸ1.1-kb nafY transcript. The position and sizes of some molecular mass markers are indicated on the left. The lower panel shows the ethidium bromide staining of total RNA in the samples, carried out as a loading control.  b Acetylene reduction assays were carried out for 15 min at 30°C after titration with an excess of dinitrogenase reductase as described (22). c Stressing conditions refer to cells grown at 37°C and molybdenum stress. extracts of A. vinelandii strains grown under stressing nitrogen fixation conditions did not increase the levels of nitrogenase activity already present in the extracts (compare with values in Table II). The lack of activation by FeMo-co was not fixed by the addition of dinitrogenase reductase to the extract during FeMo-co insertion. Again, nafY mutants showed very small levels of dinitrogenase activity compared with wild type, indicating that nafY mutants do not accumulate substantial amounts of FeMo-co-activable apodinitrogenase in stressing conditions, but they rather have a very unstable form of dinitrogenase.
Taken together these results are consistent with a model in which NafY and NifX, but not NifY, are capable of playing a role in either insertion of FeMo-co into apodinitrogenase or stabilization of apodinitrogenase in an open activable conformation. These two possible roles are likely to be very important under natural environmental conditions, where limiting amounts of molybdenum are available and apodinitrogenase must be maintained and protected in an open (FeMo-co-deficient) conformation. Although we cannot conclusively rule out a role for NafY in the stabilization of FeMo-co inside the cell, no substantial dinitrogenase reactivation is observed when purified FeMo-co is added to cell extracts of nafY mutants grown under stressing conditions.
In Vitro Activation of Apodinitrogenase in Extracts of Strains UW45 and UW146 by Purified FeMo-co-Because of its ability to bind independently to FeMo-co and to apodinitrogenase, both FeMo-co insertase and chaperone roles have been proposed for the ␥ protein (13). A tetrameric (␣ 2 ␤ 2 ) His-tagged apodinitrogenase is, however, competent for FeMo-co activation despite lacking ␥ (14). We have compared the FeMo-co-dependent activation properties of ␥-deficient (␣ 2 ␤ 2 ?) and hexameric (␣ 2 ␤ 2 ␥ 2 ) apodinitrogenases present in extracts of strains UW146 (nifBnafY) and UW45 (nifB), respectively. In this work, we will use a question mark in the description of the ␥-deficient apodinitrogenase because we do not yet know its exact subunit composition. To demonstrate that the mutation in nafY is not affecting the accumulation of apodinitrogenase in the extracts of UW146, the same amount of total protein from extracts of UW45 and UW146 was resolved by SDS-PAGE and analyzed by immunoblot, developed with antibodies to dinitrogenase (Fig. 5). The amount of anti-NifDK-reacting material in the immunoblot was then quantified by densitometric analysis and shown as relative numbers in Fig. 5. These data indicate that under standard nitrogen fixation conditions, the nafY mutation does not result in decreased levels of apodinitrogenase in extracts of UW146.
30 min after the addition of FeMo-co, apodinitrogenase lacking ␥ (UW146) could only be activated to a value of 9.2 nmol of ethylene formed/min/mg of protein, which is 57% of the value seen for the ␥-containing apodinitrogenase from UW45 (see Table IV). UW146 and UW45 extracts contained similar levels of apodinitrogenase (see above) and dinitrogenase reductase activities (23.5 Ϯ 2.8 nmol of ethylene produced/min/mg of protein). These data considered, the evident difference in activation could only be caused either by a lower insertase activity or a lower level of activable apodinitrogenase in extracts of strain UW146. These results are in contrast to the case in K. pneumoniae, where a nifBnifY mutant is severely affected in the levels of FeMo-co-activable apodinitrogenase (10) and in apodinitrogenase antigen as well. 3 Fig. 6 illustrates a time course of FeMo-co insertion into apodinitrogenase over a period of 30 min. Apodinitrogenase in extracts of UW45 and UW146 presented a similar activation profile, except that the maximum level of activation is halved in extracts of UW146. It is clear that the lack of ␥ in extracts of UW146 does not limit the rate of FeMo-co insertion during the reaction. Moreover, no enhancement of FeMo-co insertion was obtained by combining extracts of nif-derepressed UW146 cells and ammonium-grown wild-type cells containing ␥ protein, as expected if ␥ were involved in the insertion reaction (data not shown). Thus, the addition of ␥ to the in vitro insertion reaction mixture is not enough to restore the levels of activable apodinitrogenase in UW146 extracts. Results presented in Table IV and Fig. 6 indicate that a lower level of activable apodinitrogenase, and not a defect in FeMo-co insertion, is the main cause of reduced holodinitrogenase reconstitution in extracts of strain UW146. It is very interesting that the value of holodinitrogenase formation from ␥-free apodinitrogenase is about 50% of the value obtained from the ␥-containing enzyme. Experiments to distinguish whether there is only one nonactivable FeMo-co pocket per ␣ 2 ␤ 2 apodinitrogenase tetramer or if half of its population is nonactivable by FeMo-co are currently being developed. 3 M. J. Homer and G. P. Roberts, unpublished results. The FeMo-co activation assay system contained 0.2 ml of extract (0.6 -1.4 mg of protein) and purified FeMo-co in N-methylformamide. The reaction mixtures were incubated at 30°C for 30 min, after which 0.8 ml of a MgATP-regenerating system and an excess of purified dinitrogenase reductase (0.2 mg of protein) were added, and the acetylene reduction activity was assayed.
b Dinitrogenase reductase (0.2 mg of protein) was added to the FeMo-co insertion phase and to the activity phase. Values are the averages of at least two assays performed separately. Activities are expressed as nmol of ethylene formed/min/mg of protein.  a The FeMo-co activation assay system contained 0.2 ml of extract from UW45 or UW146 strain (Ϸ3.0 mg of protein) and purified FeMo-co in N-methylformamide. The reaction mixtures were incubated at 30°C for 30 min, after which 0.8 ml of a MgATP-regenerating system and an excess of purified dinitrogenase reductase (0.2 mg of protein) were added, and the acetylene reduction activity was assayed (see "Experimental Procedures").
b Activity is expressed as nmol of ethylene formed/min/mg of protein. Values are the averages of at least five assays.
FIG. 6. Time course of apodinitrogenase activation by purified FeMo-co in extracts of UW45 (nifB) (q) and UW146 (nifBnafY) (E) strains. The reaction mixture contained 0.7 ml of Tris⅐HCl buffer, 1.4 ml of UW45 or UW146 cell-free extracts (Ϸ21 mg of protein), and purified FeMo-co in N-methylformamide (equivalent to 3.5 nmol of molybdenum). At the times indicated, 0.2-ml aliquots (Ϸ2 mg of protein) were removed from the mixture and treated as described under "Experimental Procedures." Data shown are representative of a number of experiments, which showed very similar results.