Incorporation of iron and sulfur from NifB cofactor into the iron-molybdenum cofactor of dinitrogenase.

NifB-co is an iron- and sulfur-containing precursor to the iron-molybdenum cofactor (FeMo-co) of dinitrogenase. The synthesis of NifB-co requires at least the product of the nifB gene. Incorporation of 55Fe and 35S from NifB-co into FeMo-co was observed only when all components of the in vitro FeMo-co synthesis system were present. Incorporation of iron and sulfur from NifB-co into dinitrogenase was not observed in control experiments in which the apodinitrogenase (lacking FeMo-co) was initially activated with purified, unlabeled FeMo-co or in assays where NifB-co was oxygen-inactivated prior to addition to the synthesis system. These data clearly demonstrate that iron and sulfur from active NifB-co are specifically incorporated into FeMo-co of dinitrogenase and provide direct biochemical identification of an iron-sulfur precursor of FeMo-co. Under different in vitro FeMo-co synthesis conditions, iron and sulfur from NifB-co were associated with at least two other proteins (NIFNE and gamma) that are involved in the formation of active dinitrogenase. The results presented here suggest that multiple FeMo-co processing steps might occur on NIFNE and that FeMo-co synthesis is most likely completed prior to the association of FeMo-co with gamma.

NifB-co is an iron-and sulfur-containing precursor to the iron-molybdenum cofactor (FeMo-co) of dinitrogenase. The synthesis of NifB-co requires at least the product of the nifB gene. Incorporation of 55 Fe and 35 S from NifB-co into FeMo-co was observed only when all components of the in vitro FeMo-co synthesis system were present. Incorporation of iron and sulfur from NifB-co into dinitrogenase was not observed in control experiments in which the apodinitrogenase (lacking FeMo-co) was initially activated with purified, unlabeled FeMo-co or in assays where NifB-co was oxygen-inactivated prior to addition to the synthesis system. These data clearly demonstrate that iron and sulfur from active NifB-co are specifically incorporated into FeMo-co of dinitrogenase and provide direct biochemical identification of an ironsulfur precursor of FeMo-co.
Under different in vitro FeMo-co synthesis conditions, iron and sulfur from NifB-co were associated with at least two other proteins (NIFNE and gamma) that are involved in the formation of active dinitrogenase. The results presented here suggest that multiple FeMo-co processing steps might occur on NIFNE and that FeMo-co synthesis is most likely completed prior to the association of FeMo-co with gamma.
The conversion of dinitrogen to ammonium by biological systems is catalyzed by nitrogenase. Nitrogenase is composed of two oxygen-labile metalloproteins: dinitrogenase (also called MoFe protein or component I) and dinitrogenase reductase (also called NIFH, Fe protein, or component II; Refs. 1 and 2). Dinitrogenase is an ␣ 2 ␤ 2 tetramer of the nifD and nifK gene products, and it contains two pairs of unique metal clusters, known as the iron-molybdenum cofactor (FeMo-co; 1 Refs. 3 and 4) and the P-cluster (4,5). Dinitrogenase is specifically reduced by dinitrogenase reductase. Dinitrogenase reductase, which contains a single Fe 4 S 4 cluster, is an ␣ 2 dimer of the nifH gene product (6). The electrons transferred to dinitrogenase are ultimately channeled to FeMo-co, the site of substrate reduction (see Ref. 7

for a concise review).
FeMo-co is composed of molybdenum, iron, sulfur, and homocitrate ((R)-2-hydroxyl-1,2,4-butane tricarboxylic acid) in a ratio of 1:7:9:1 (4,8). The products of at least six nitrogen fixation (nif) genes, including nifQ, nifV, nifB, nifH, nifN, and nifE, are required for the biosynthesis of FeMo-co (8 -11). Interestingly, 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 FeMo-co-deficient dinitrogenase (apodinitrogenase; Refs. 12 and 13). The high degree of sequence similarity between the nifN and nifK sequences and the nifE and nifD sequences suggests that NIFNE might serve as a scaffold for FeMo-co biosynthesis (14). This hypothesis is supported by the recent observation that the mobility of NIFNE on native (nondenaturing) gels changes specifically upon the addition of NifBco, a likely FeMo-co precursor (described below; Ref. 15). An in vitro FeMo-co synthesis system that requires an ATP-regenerating system, molybdate, homocitrate, and at least NIFB, NIFNE, and NIFH has been described (11). Although use of the in vitro system has yielded significant information concerning FeMo-co biosynthesis, the nature of the iron and sulfur donor(s) for the biosynthesis of FeMo-co remains unknown.
NifB-co is one potential source of iron and sulfur for FeMo-co biosynthesis. In the course of attempting to purify the NIFB protein from Klebsiella pneumoniae, Shah et al. (16) isolated and purified the apparent product of NIFB as a detergentsolubilized, small molecule termed NifB-cofactor (NifB-co). Solutions of NifB-co exhibit certain characteristics that are similar to solutions of purified FeMo-co, including color, stability in N-methylformamide, and oxygen lability. The requirement for NIFB in the in vitro FeMo-co synthesis assay is satisfied by the addition of NifB-co, and the amount of FeMo-co synthesized in vitro is proportional to the amount of NifB-co added to the system in which all other components are present in excess. The stoichiometric requirement of NifB-co is consistent with the hypothesis that NifB-co is an iron-sulfur precursor of FeMo-co. Because a functional nifB gene is also required for the molybdenum-independent nitrogen fixation systems (17), it has been proposed that NifB-co is the basic iron-sulfur cluster for the synthesis of FeMo-co, the vanadium-containing cofactor (FeV-co) of the vnf-encoded nitrogenase, and the iron-only cofactor of the anf-encoded nitrogenase (16).
In vitro activation of apodinitrogenase by FeMo-co apparently requires the presence of a protein designated as gamma (18). Recent studies show that addition of purified FeMo-co to crude extracts and partially purified fractions containing gamma results in a shift in the electrophoretic mobility of gamma on native gels. The mobility change correlates with the incorporation of iron into gamma (19). In addition, crude extracts that contain this faster migrating form of gamma (with associated FeMo-co) are capable of activating apodinitrogenase in vitro (19).
To date there has been no direct evidence for the incorporation of iron and sulfur from NifB-co into FeMo-co or for the flow of iron and sulfur from NifB-co through the NIFNE and gamma proteins. This report directly demonstrates that iron and sulfur from NifB-co can become associated with NIFNE and gamma, and ultimately become incorporated into apodinitrogenase as FeMo-co.  20), and DJ678 (⌬nifDK ⌬YENX::kan) were grown, derepressed, and extracts were prepared as described (3). Strain DJ678 (⌬nifDK ⌬YENX::kan) was constructed by transforming strain DJ33 (21) with pDB583. Plasmid pDB583 was derived from pDB42 by replacing an internal BglII fragment with a 1.6-kilobase BamHI fragment that carries a kanamycin resistance cartridge from pUC4-KIXX (from Pharmacia). DJ678 resulted from a double-reciprocal recombination event where the deletion and insertion mutations located in pDB583 were transferred to the chromosome. Plasmid pDB583 is not capable of autonomous replication in A. vinelandii. Experimental details of similar strain constructions are described in Ref. 21.

Materials-Sephacryl
Strain UW45 was grown and derepressed in tungsten-containing medium (molybdenum-free) as described previously (11). When necessary, small molecules (i.e. homocitrate) were removed from the cell extracts by Sephadex G-25 column chromatography.
K. pneumoniae Strain and Growth Conditions-K. pneumoniae strain UN1217 (nifN4536) has been described (22). The minimal medium (23) was modified as described previously (16). When derepressing the culture in the presence of Na 2 35 SO 4 , 2.5 g of MgCl 2 ⅐6H 2 O was substituted for MgSO 4 ⅐7H 2 O (the only source of sulfur in the medium). FeCl 3 was excluded from medium used for derepression of strain UN1217 in the presence of 55 FeCl 3 . Contaminating metals were removed from the phosphate salts by passage over a Chelex-100 column and all glassware was rinsed overnight with 4 N HCl.
For sulfur donation studies, a starter culture of strain UN1217 was grown aerobically on a rotary shaker at 30°C for 20 -24 h in 275 ml of medium (minimal medium as described previously; Ref. 16) containing 15.6 mM ammonium acetate (filter-sterilized). One ml of this culture (stationary phase) was transferred to 250 ml of fresh medium containing 0.2 mM sulfur added as Na 2 SO 4 . Approximately 15 h later, this starter culture was used to inoculate a larger culture that would be derepressed. A 5-liter Pyrex carboy containing 4.5 liter of the medium plus 3 ml of a 22% ammonium acetate solution (approximately 1.9 mM final concentration) and 28.7 ml of a 10 mM Na 2 SO 4 solution (approximately 0.063 mM sulfur final concentration) was inoculated with 105 ml of the starter culture. The large culture was grown at 30°C and sparged vigorously with compressed air (filter-sterilized). The culture was spot tested for ammonium using Nessler's reagent. Thirty min after exhaustion of the ammonium (which occurred approximately 4 h post inoculation), derepression of the Nif proteins was initiated by switching the gas from air to 95% N 2 ϩ 5% CO 2 and by adding 4.4 ml of a sterile 10% L-serine solution. At this time, approximately 4 mCi of Na 2 35 SO 4 (specific activity 561-572 mCi/mmol) were added to the culture. Six hours after the start of derepression, the cells were harvested (approximately 5 g, wet weight) by centrifugation and stored at Ϫ80°C. Approximately 60% of the radiolabel added to the medium was incorporated into the cells.
Growth and derepression of strain UN1217 in the presence of 55 FeCl 3 were similar to that described above, with the following modifications. The starter cultures were not supplemented with any iron. The medium in the 5-liter carboy was supplemented with FeCl 3 to a final concentration of approximately 10 M. At the start of derepression, 2 mCi of 55 FeCl 3 (specific activity 26.7-29.2 mCi/mg) were added. 55 Fe-Labeled ferric citrate was prepared by diluting the 55 FeCl 3 into 1 ml of distilled H 2 O containing 5 mg of sodium citrate (pH approximately 7). The solution was heated and the entire volume was added to the carboy. Approximately 50 -70% of the radiolabel added to the medium was incorporated into the cells.
UN1217 Cell Breakage and Preparation of Crude NifB-co Samples-All buffers used throughout the preparation of the crude extracts containing NifB-co were deoxygenated as described previously and contained 0.2 mM phenylmethylsulfonyl fluoride, 0.5 g/ml leupeptin, and 1.7 mM DTH (16). Anoxic conditions were maintained throughout the procedure. The UN1217 cells were suspended in 0.1 M Tris-HCl (pH 7.4, 2 ml/g wet weight of cells) in a centrifuge tube equipped with a screwcap modified to accommodate a rubber stopper (so that the tube could be flushed with argon on a gassing manifold). Deoxyribonuclease (approximately 1 mg/15 ml), ribonuclease (approximately 1 mg/15 ml), and lysozyme (approximately 3 mg/15 ml) were added to the suspension. The suspension was incubated at 30°C for 1 h. Approximately 4.5 ml of the cell suspension were transferred to 10-ml polycarbonate centrifuge tubes (16 ϫ 76 mm) that contained 9.5 g of 0.1-0.2 mm glass beads. The beads had been equilibrated overnight with anoxic 0.1 M Tris-HCl buffer (pH 7.4). The tubes were degassed, and cell breakage was initiated by vortexing (standard bench-top vortexer on high setting) each tube for 1 min. Each tube was placed on ice for at least 2 min following vortexing. Seven vortex/ice cycles were used for K. pneumoniae cell breakage. N-Lauroylsarcosine was added to a final concentration of 2% (from a 20% anoxic stock solution in 0.1 M Tris-HCl (pH 7.4)). After 10 min at room temperature, the tubes were centrifuged at 27,200 ϫ g for 10 min. The temperature during centrifugation was maintained at approximately 20°C to avoid precipitation of the detergent. The supernatants containing NifB-co were anoxically transferred and combined. To ensure that all of the NifB-co was collected, 2 ml of 0.1 M Tris-HCl (pH 7.4) and 0.2 ml of 20% N-lauroylsarcosine stock solution were added to each tube immediately after the supernatant was removed. The tubes were degassed, vortexed for 1 min, and centrifuged as described above. The supernatants were anoxically transferred and combined (but kept separately from the first supernatant). Both supernatants were assayed for NifB-co activity using the in vitro FeMo-co synthesis assay (described below). The supernatants were stored at Ϫ20°C.
NifB-co Purification-All buffers used throughout the purification of NifB-co contained 1.7 mM DTH. The NifB-co was purified according to the published protocol with some key modifications to accommodate small scale purification of radiolabeled NifB-co (16). The supernatants described above were combined and diluted 3-fold into 0.1 M Tris-HCl (pH 7.4). DTT was added to a final concentration of 5 mM, and the diluted sample was incubated for 1 h at room temperature. The preparation was applied to a 1 ϫ 8.5-cm Sephacryl S-200 column equilibrated with 0.1 M NaCl in 25 mM MOPS (pH 7.5) containing 1 mM DTT and 0.4% N-lauroylsarcosine. NifB-co binds to Sephacryl S-200 under these conditions. The column was washed with two column volumes of the equilibration buffer and the NifB-co was eluted with 0.1 M NaCl in 25 mM MOPS (pH 7.5) containing 1 mM DTT and 2% SB-12. The column flow rate during the load and elution was approximately 19 ml/h. Fractions were collected anoxically and assayed for NifB-co activity using the in vitro FeMo-co synthesis system (described below). 35 S-Labeled and 55 Fe-labeled NifB-co were successfully purified from 4 and 3 different lots of cells, respectively.
In Vitro FeMo-co Synthesis Assay-Nine-ml serum vials were flushed with purified argon and rinsed with anoxic 25 mM Tris-HCl (pH 7.4). A complete FeMo-co synthesis reaction mixture was prepared by combining the following: 100 l of 25 mM Tris-HCl (pH 7.4), 10 l of 1 mM Na 2 MoO 4 , 20 l of 5 mM homocitrate (that had been treated with base to cleave the lactone, pH 8.0), and 200 l of an 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 DTH). The reaction mixtures were incubated at room temperature for 10 -15 min. Two hundred l of the appropriate A. vinelandii cell-free extract (except in FeMo-co activated samples), 25 l of the Sephacryl S-200 fraction containing NifB-co, and 10 l of purified dinitrogenase reductase (0.1 mg of protein) were added to the reaction mixtures. The vials were incubated in a rotary water-bath shaker at 30°C for 35 min. After this incubation, samples to be applied onto native, polyacrylamide gels were placed on ice. The activity of the newly formed dinitrogenase was measured in the remainder of the vials using the C 2 H 2 reduction assay as described previously (11).
In control experiments, various components of the complete reaction mixture were excluded as indicated under "Results and Discussion." Where indicated, apodinitrogenase in 200 l of the appropriate extract was activated by incubation with an excess of purified, unlabeled FeMo-co for 10 min before performing the in vitro FeMo-co synthesis reaction with the labeled NifB-co.
Activation of Apodinitrogenase by FeMo-co (FeMo-co Insertion Assay)-The preparation of FeMo-co and the FeMo-co insertion assays were performed as described previously (3). Where indicated, NifB-co was added to the reaction mixture and incubated for 20 min prior to the addition of FeMo-co.
Anoxic Native Gel Electrophoresis-One hundred l of anoxic 50% glycerol were added to reaction mixtures that would be applied onto the gels. Unless otherwise indicated, 100 l of the reaction mixtures (with added glycerol) were applied onto the gels. Proteins were separated on anoxic, native gels with a 5-10% acrylamide (37.5 acrylamide:1 bisacrylamide) and 0 -20% sucrose gradient and 400 mM Tris-HCl buffer (pH 9.0). Gel dimensions were approximately 14 ϫ 10 ϫ 0.15-cm. The reservoir buffer was N 2 -sparged 65 mM Tris-glycine (pH 8.6) containing 1.7 mM DTH. Gels were prerun for at least 45 min at 122 V for initial reduction and proteins were electrophoresed for 2,300 V ϫ h (approximately 122 V) or 900 V ϫ h (approximately 60 V). Gels were run at 4°C.
Immunoblots-The protocols for immunoblotting and developing (24) with modifications by Brandner et al. (25) have been described. Native gels were equilibrated in the transfer buffer for at least 15 min before blotting. Polyclonal antibodies to dinitrogenase were raised in rabbits.
Visualization of Radioactivity-Gels were exposed to a phosphor screen for 1-3 days and were scanned using a Molecular Dynamics model 425e PhosphorImager.
Acid-labile Sulfide Determination-The method of Brumby et al. (26) was used to confirm that acid-labile sulfide is a component of NifB-co. The tubes used for analysis were stoppered and placed under an argon atmosphere on a gassing manifold. All manipulations were done using a Hamilton syringe inserted through the stopper.

RESULTS AND DISCUSSION
Identification of an Iron and Sulfur Donor for FeMo-co Synthesis-In the initial study, purified NifB-co was reported to contain iron as the only metal (16). The similarities between solutions of NifB-co and FeMo-co suggested that NifB-co was an iron-sulfur cluster, although the presence of acid-labile sulfide was not demonstrated. Prior to investigating the role of NifB-co as a sulfur donor for FeMo-co synthesis, the presence of acid-labile sulfide in NifB-co was confirmed (data not shown). Various components of the buffer used to purify NifB-co, including N-lauroylsarcosine, DTT, and DTH, interfered with the colorimetric assay for acid-labile sulfide, and therefore precise quantitation of the iron:sulfur ratio of NifB-co was not possible. The data revealed, however, that purified NifB-co preparations contained significant amounts of acid-labile sulfide. Modifications of the NifB-co purification scheme and of the assay for acid-labile sulfide are currently being investigated to accurately quantitate the iron and sulfur content of NifB-co.
To investigate the role of NifB-co as an iron-sulfur donor for FeMo-co biosynthesis, NifB-co was independently labeled in vivo with 55 Fe or 35 S. The labeled cofactor was purified as described under "Experimental Procedures." The 55 Fe-and 35 Slabeled NifB-co fractions contained an average of 2,200,000 and 300,000 cpm/ml, respectively. The average activities of the 55 Fe and 35 S-labeled NifB-co fractions were 770 and 400 nmol of C 2 H 2 reduced/min by dinitrogenase formed/ml of the NifB-cocontaining fraction, respectively.
The in vitro FeMo-co synthesis system together with anoxic, native gel electrophoresis was employed to monitor the incorporation of iron and/or sulfur from NifB-co into the FeMo-co of dinitrogenase. A complete reaction mixture that included all of the components known to be required for FeMo-co biosynthesis was used to monitor donation of iron and sulfur from NifB-co to FeMo-co. The complete reaction mixtures contained molybdenum, homocitrate, an ATP-regenerating system, and cell-free extract from strain UW45 (nifB), which served as a source of NIFNE, dinitrogenase reductase, apodinitrogenase, and any other unidentified factors required for in vitro FeMo-co biosynthesis. Purified 55 Fe-or 35 S-labeled NifB-co was added to complete the reaction mixture. A number of control reactions (in which iron and sulfur from NifB-co were not expected to be incorporated into dinitrogenase) were performed to demonstrate the specificity of incorporation of radiolabel from active NifB-co into dinitrogenase. Homocitrate, molybdenum, and MgATP were omitted from some in vitro FeMo-co synthesis reaction mixtures to prevent FeMo-co synthesis. In other control reaction mixtures, the oxygen-labile, labeled NifB-co was inactivated by exposure to air prior to addition to the reactions. In other control assays, all of the apodinitrogenase present in the UW45 extracts was activated in vitro with unlabeled, purified FeMo-co prior to the addition of labeled NifB-co to the complete reaction mixture. In this system, all of the available FeMo-co binding sites on the apodinitrogenase should be occupied by unlabeled FeMo-co, and therefore iron and sulfur from labeled NifB-co were not expected to be associated with dinitrogenase.
To investigate the incorporation of 55 Fe and 35 S from NifB-co into FeMo-co, the proteins in the various in vitro FeMo-co synthesis reaction mixtures were separated on anoxic native gels. The position to which dinitrogenase migrated was determined by immunoblot analysis (data not shown). The data in Fig. 1A revealed that incorporation of 55 Fe from labeled NifB-co into dinitrogenase required the presence of all of the components known to be required for in vitro FeMo-co synthesis. A prominently labeled band that comigrated with dinitrogenase was only detected in the lanes to which the complete FeMo-co synthesis reaction mixture (plus apodinitrogenase) was applied (Fig. 1A, lanes 7 and 8). The species that migrated slightly faster than dinitrogenase has been identified as NIFNE and is discussed in detail below. At least five lines of evidence suggested that in the complete in vitro FeMo-co synthesis reaction mixture, iron from NifB-co was specifically incorporated into the FeMo-co of dinitrogenase. (i) Only very low levels of iron were associated with dinitrogenase when FeMo-co synthesis was prevented in the reaction mixtures by omission of MgATP, molybdenum, or homocitrate (each of which is a known requirement for in vitro FeMo-co synthesis; see Fig. 1A, lanes 2, 6, and 9, respectively). (ii) The absence of a band that co-migrated with dinitrogenase in samples where the air-inactivated 55 Fe-NifB-co was utilized demonstrated that active NifB-co was required for incorporation of the 55 Fe into FeMo-co of dinitrogenase (Fig. 1A, lane 5). (iii) Activation of apodinitrogenase with unlabeled FeMo-co prior to synthesizing FeMo-co using the 55 Fe-labeled NifB-co resulted in almost no association of 55 Fe with dinitrogenase (Fig. 1A, lane 3). (iv) No radiolabel was detected at the dinitrogenase position in the lane containing only free 55 Fe-NifB-co (Fig. 1A, lane 4). (v) Labeled dinitrogenase was not observed when the complete reaction mixture was oxidized following the FeMo-co synthesis reaction, but prior to being applied to the gel (data not shown). Note that the C 2 H 2 reduction activities of the various assays (Fig. 1A) are consistent with the conclusion that holodinitrogenase was only formed in the complete system (by FeMo-co synthesis using 55 Fe-NifBco; Fig. 1A, lanes 7 and 8) and in the FeMo-co activated sample (by activation of apodinitrogenase with unlabeled FeMo-co; Fig. 1A, lane 3). Similar results were obtained when 35 S-labeled NifB-co was used in the various in vitro FeMo-co synthesis reaction mixtures (Fig. 1B, compare lanes 1, 2, and 4 (control experiments) with lane 3 (complete system)). The differences in the band intensities in reactions that used 35 S-labeled NifB-co compared to those that used 55 Fe-labeled NifB-co (Fig. 1, compare A and B) are most likely because the cpm/ml of the 35 S-labeled NifB-co was only 14% that of the 55 Fe-labeled NifBco. Together these data show that iron and sulfur from NifB-co were only incorporated into dinitrogenase under conditions where FeMo-co was synthesized and demonstrate that iron and sulfur from NifB-co were specifically incorporated into the FeMo-co of dinitrogenase. These data provide direct biochemical identification of an iron and sulfur precursor of FeMo-co.
The lack of 55 Fe associated with dinitrogenase in the sample in which apodinitrogenase was initially activated with purified, unlabeled FeMo-co suggested that, once bound to the protein, there was not a significant amount of turnover of FeMo-co in this system. The absence of a labeled dinitrogenase band indicated that the synthesized 55 Fe-FeMo-co did not displace the FeMo-co that originally activated the apodinitrogenase. The failure of NifB-co to form a complex with apodinitrogenase was also tested by investigating its possible ability to inhibit the insertion of FeMo-co into apodinitrogenase. As shown in Table I, preincubation of apodinitrogenase with an excess of NifB-co did not result in any detectable inhibition of FeMo-co insertion, consistent with the labeling results in Fig. 1.
Accurate quantitation of the number of iron and sulfur atoms donated by NifB-co for FeMo-co synthesis remains to be accom-plished. However, based on previous activity and iron analysis studies, NifB-co has been predicted to donate all of the iron for FeMo-co biosynthesis (16).
In the complete in vitro FeMo-co synthesis reaction mixture (Fig. 1A, lanes 7 and 8), the iron from NifB-co was obviously associated with more than one protein. The slowest migrating species is currently unidentified (Fig. 1A, lane 8). However, the species was also present in extracts of wild type cells grown on NH 4 ϩ (data not shown) and was likely binding iron from denatured NifB-co because it was also observed in samples containing oxidized NifB-co (Fig. 1A, lane 5) and, in fact, became more prominent as the labeled NifB-co samples lost activity, most likely due to oxygen inactivation. The slowest migrating species was not detected when 35 S-labeled NifB-co was used (Fig. 1B).
Iron and Sulfur from NifB-co Associate with NIFNE-NIFNE has been proposed to serve as a scaffold for FeMo-co biosynthesis (14,15). This hypothesis is based on several lines of evidence. There is a high degree of sequence similarity between the nifN and nifK sequences and the nifE and nifD sequences. Most notably, Cys-275 of NIFD, which serves as a ligand to FeMo-co in the dinitrogenase protein, is conserved in NIFNE (14). In addition, the mobility of NIFNE on native gels has been shown to respond specifically to the addition of NifB-co (15). NIFB activity has been shown to initially copurify with NIFNE in extracts from certain mutant backgrounds. It has been suggested that this reflects a direct interaction between NifB-co and NIFNE (15). It was therefore of interest to specifically address whether iron and sulfur from NifB-co could associate with NIFNE under various in vitro FeMo-co synthesis reaction conditions. The labeled species that migrated slightly faster than dinitrogenase (Fig. 1A, lanes 7 and 8) was identified as NIFNE by immunoblot analysis (data not shown) and by use of purified NIFNE protein (Fig. 1A, lane 1). Purified NIFNE with associated 55 Fe from NifB-co (Fig. 1A, lane 1) comigrated with the labeled species observed in a number of the in vitro FeMo-co synthesis reaction mixtures including: (i) reaction mixtures from which MgATP (Fig. 1A, lane 2), molybdenum (Fig. 1A,  lane 6), homocitrate (Fig. 1A, lane 9), or dinitrogenase reductase (data not shown) were excluded, (ii) complete in vitro FeMo-co synthesis reaction mixtures in which all of the apodinitrogenase was activated with unlabeled FeMo-co prior to the addition of the 55 Fe-NifB-co (Fig. 1A, lane 3), and (iii) the complete reaction mixture (Fig. 1A, lanes 7 and 8). Active NifB-co was required for this association, as determined by the absence of this species when air-inactivated 55 Fe-NifB-co was

TABLE I NifB-co does not inhibit the insertion of FeMo-co into apodinitrogenase
FeMo-co insertion system a Activity b UW45 ϩ FeMo-co 14.5 UW45 ϩ NifB-co c ϩ FeMo-co 14.4 UW45 ϩ inactivated NifB-co d ϩ FeMo-co 14.2 a FeMo-co insertion assays contained 0.1 ml of 25 mM Tris-HCl (pH 7.4) and 0.1 ml of desalted UW45 extract as a source of apodinitrogenase (1.5 mg of protein). Where indicated, 20 l of purified NifB-co (in buffer containing 2% N-lauroylsarcosine) was added and the mixtures were incubated for 20 min at room temperature. Purified FeMo-co (in NMF) was then added. After further incubation, 0.8 ml of ATP-regenerating mixture and an excess of purified dinitrogenase reductase (0.1 mg) were added and the C 2 H 2 reduction activities were monitored.
b Expressed as nanomoles of C 2 H 4 formed/min/assay. c In the in vitro FeMo-co synthesis assay (with all other components in excess), 20 l of NifB-co supported a dinitrogenase activity of 20 nmol of ethylene formed/min/assay. d NifB-co was oxidized in air for Ͼ15 min prior to addition to the assay. used in the FeMo-co synthesis reactions (Fig. 1A, lane 5). These data suggest that in the absence of molybdenum, homocitrate, MgATP, or dinitrogenase reductase (each of which is a requirement for FeMo-co synthesis) NifB-co is associated with NIFNE. These results are consistent with a model where these compounds are all necessary for later steps of FeMo-co synthesis on NIFNE and suggest that FeMo-co might be completed on NIFNE. The 55 Fe-NifB-co data presented here definitively demonstrate that the previously observed shift in the mobility of NIFNE on native gels in the presence of NifB-co was due to the association of iron and sulfur from NifB-co with NIFNE.
It is currently unclear why less 55 Fe from NifB-co was reproducibly associated with NIFNE when homocitrate was excluded from the reaction mix compared to when MgATP or molybdenum were excluded (Fig. 1A, compare lane 9 with lanes 2 and 6). At least two hypotheses might explain this observation. Although the activity assay showed no C 2 H 2 reduction activity (which suggests that no FeMo-co was synthesized), an organic acid other than homocitrate might have been present in the reaction mixture, and thus a low amount of an aberrant form of FeMo-co (containing an organic acid other than homocitrate) might have been synthesized and proceeded along the biosynthetic pathway (i.e. been passed to another protein). The ability of other organic acids to substitute for homocitrate in vitro has been demonstrated (27). Alternatively, it is possible that in this reaction system (that lacked homocitrate) the ironsulfur precursor was converted to a form (possibly with added molybdenum, but not the finished FeMo-co) that had a lesser affinity for NIFNE than did the iron-sulfur precursor.
Binding of 55 Fe and 35 S from NifB-co to NIFNE was also observed in reactions that utilized cell extracts of strains with mutations in nifDK (apodinitrogenase mutants; CA11.1 (⌬nif-HDK ⌬vnfDGK1::spc), which produces both NIFNE and NifB-co (Fig. 2, lane 2); DJ677 (⌬nifB::kan ⌬nifKD), which produces NIFNE (Fig. 1B, lane 5)). Interestingly, the observed amount of 55 Fe from NifB-co associated with NIFNE in extracts of strain CA11.1 in the presence and absence of added dinitrogenase reductase suggests that the NIFNE-NifB-co complex is a physiologically relevant precursor along the FeMo-co biosynthetic pathway. Significantly more 55 Fe (from NifB-co) was associated with NIFNE from CA11.1 (⌬nifHDK ⌬vnfDGK1::spc) extracts when dinitrogenase reductase was added in addition to all of the other requirements for in vitro FeMo-co synthesis (Fig. 2, compare lanes 2 and 3). Extracts of strain CA11.1 derepressed for the Nif proteins exhibited high levels of NIFB and NIFNE activities as monitored by the FeMo-co synthesis assay. Immunoblot analysis of a native gel containing crude extract from this strain showed that the NIFNE present migrated at the position for NIFNE with bound NifB-co (data not shown; Ref. 15). We hypothesize that the majority of NIFNE in lane 3 (Fig. 2) had bound NifB-co (unlabeled, of in vivo origin), and therefore little association of the added 55 Fe-NifB-co with NIFNE was observed in the absence of a complete in vitro FeMo-co synthesis reaction system. When dinitrogenase reductase was added to complete the reaction mixture containing CA11.1 extract, molybdenum, homocitrate, and MgATP, all of the components known to be required for FeMo-co biosynthesis were present and therefore FeMo-co was synthesized. Thus, the unlabeled NifB-co (of in vivo origin) was apparently incorporated into FeMo-co and proceeded along the biosynthetic pathway (discussed below). The NIFNE protein was then available to interact with the 55 Fe-NifB-co that had been added to the system, and therefore an increase in 55 Fe associated with NIFNE was observed upon addition of dinitrogenase reductase to the reaction mixture (Fig. 2, lane 2). These results demonstrate the relevance of the interaction of NifB-co with NIFNE. The apparent ability to chase NifB-co from NIFNE in CA11.1 extracts by addition of dinitrogenase reductase to the reaction mixture suggests that NIFNE is not only capable of binding NifB-co, but, in fact, the NIFNE-NifB-co complex appears to be a precursor of FeMo-co.
It is currently unclear if FeMo-co synthesis is completed on NIFNE, however there is evidence that NIFNE can bind to FeMo-co in crude extracts. When purified FeMo-co was added to extracts of strain DJ677 (⌬nifB::kan ⌬nifKD), which lacks apodinitrogenase and is unable to synthesize NifB-co, and the mixture was applied to a Sephacryl S-200 sizing column, the NIFNE-containing fraction contained FeMo-co as determined by the ability to activate apodinitrogenase in extracts of strain UW45 (nifB; data not shown). When a similar experiment was done by adding purified FeMo-co to extracts of strain DJ678 (⌬nifDK ⌬YENX::kan), which lacks both apodinitrogenase and NIFNE, no FeMo-co was associated with the Sephacryl S-200 column fraction that corresponded to the NIFNE-containing fraction in the DJ677 experiment (data not shown). These data suggest that NIFNE is capable of associating with the completed FeMo-co molecule and are consistent with the hypothesis that FeMo-co is completed on NIFNE. The observation that 55 Fe from NifB-co was associated with NIFNE in extracts of strain UW45 that had been activated with purifed FeMo-co (Fig. 1A, lane 3) might suggest that NIFNE has a greater affinity for NifB-co than for FeMo-co.
Iron and Sulfur from NifB-co Associate with Gamma-Gamma was first identified as an additional subunit associated with purified apodinitrogenase from an A. vinelandii nifB strain (28). Subsequent studies have revealed that dinitrogenase reductase and MgATP are required to associate gamma with the apodinitrogenase which then allows in vitro activation by FeMo-co (18). Gamma might also be the 65-kDa protein-FeMo-co complex isolated by Ugalde et al. (12). Gamma has recently been shown to specifically incorporate iron upon addition of purified FeMo-co (19). It was therefore of interest to examine the association of iron and sulfur from NifB-co with gamma under a variety of FeMo-co synthesis conditions. FeMo-co synthesis assays were performed using extracts from strain DJ677 (⌬nifB::kan ⌬nifDK), which lacks apodinitrogenase and NIFB. The absence of apodinitrogenase allows the accumulation of FeMo-co on other proteins to be detected. To visualize dinitrogenase and gamma on the same system (Fig. 3), samples were electrophoresed on a gel for significantly fewer V ϫ h than the gels shown in Figs. 1 and 2. The fastest migrating species (marked "unknown" in Fig. 3) is unidentified (discussed below). The smear on the gel (that obscures 55 Fe associated with NIFNE) is attributed to 55 Fe from oxygeninactivated NifB-co (see Fig. 3, lane 1). The position of gamma (with bound FeMo-co) on the gel was determined by immunoblot analysis (data not shown).
The data in Fig. 3 revealed that significant amounts of 55 Fe from 55 Fe-NifB-co accumulated on a protein that co-migrated with gamma in extracts of strain DJ677 only when all of the components required for FeMo-co biosynthesis were present (Fig. 3, lane 5). 55 Fe was not associated with gamma when FeMo-co synthesis could not occur due to omission of the MgATP (Fig. 3, lane 3) or homocitrate (Fig. 3, lane 2) from the reaction mixture or when purified, unlabeled FeMo-co was added prior to the synthesis reaction (Fig. 3, lane 4). In addition, no label was associated with this protein when the 55 Fe-NifB-co was inactivated by exposure to air prior to addition to the FeMo-co synthesis reaction (Fig. 3, lane 1). Because all components of the in vitro FeMo-co synthesis system were required to observe 55 Fe-NifB-co-dependent labeling of gamma, these data support the conclusion that the iron associated with gamma was in the form of FeMo-co and definitively demonstrate that the iron previously observed to be associated with gamma (in the form of FeMo-co) was from NifB-co.
A 55 Fe-labeled species that migrated slightly faster than gamma was observed in some reaction mixtures (Fig. 3, lanes 2,  3, and 5). The band did not correspond to free 55 Fe-NifB-co and was dependent on active NifB-co because the species was absent in samples where air-inactivated 55 Fe-NifB-co was utilized (Fig. 3, lane 1). The species was most prominent in samples that contained all of the components required for FeMo-co synthesis (Fig. 3, lane 5) and in samples that contained all of the components required for FeMo-co synthesis except homo-citrate (Fig. 3, lane 2). Immunoblot analysis indicated that this species did not correspond to gamma or NIFNE. The possibility that this is another relevant protein involved in FeMo-co biosynthesis is currently under investigation. A similar species appears to have been observed when 99 MoO 4 2Ϫ was used to investigate incorporation of molybdenum into dinitrogenase in the presence of various homocitrate analogs (see Fig. 3 of Ref. 27).
Model of FeMo-co Biosynthesis-The results of this study suggest that the iron and sulfur from NifB-co associate with the NIFNE prior to being incorporated into FeMo-co and being transferred to gamma and ultimately to dinitrogenase. The results presented here are consistent with a biosynthetic model in which molybdenum, MgATP, and homocitrate, in addition to dinitrogenase reductase, enter the biosynthetic pathway at the level of NIFNE. Observation of the unidentified species that migrated slightly faster than gamma (and was prominently labeled with 55 Fe in a reaction mixture that lacked homocitrate), however, somewhat complicates these conclusions, and therefore definitive proof of where molybdenum and homocitrate enter the biosynthetic pathway will require use of 99 MoO 4 2Ϫ and labeled homocitrate. The gamma protein apparently associates with the completed FeMo-co molecule.
A major question that remains concerns the source of iron and sulfur for NifB-co biosynthesis. Exciting results from the laboratory of Dean and colleagues indicate that the nifU and nifS gene products are likely to be involved in the mobilization of inorganic iron and sulfur for the synthesis of the nitrogenase iron-sulfur clusters (see Ref. 7 for a recent review). A system that utilized NIFS for the reconstitution of the Fe 4 S 4 cluster of dinitrogenase reductase was recently described (29). It will be interesting to attempt to generate NifB-co using a similar system.
Conclusions--The results of this study demonstrate that iron and sulfur from NifB-co are incorporated into the FeMo-co of dinitrogenase and provide direct biochemical identification of an iron-sulfur donor for FeMo-co biosynthesis. Under different in vitro FeMo-co synthesis conditions, iron and sulfur from NifB-co were associated with at least two other proteins (NIFNE and gamma) that are involved in the formation of active dinitrogenase. The results presented here suggest that multiple FeMo-co processing steps might occur on NIFNE. The ability to "chase" NifB-co from NIFNE by altering the in vitro reaction conditions suggests that the NIFNE-NifB-co complex is a physiologically relevant precursor along the FeMo-co biosynthetic pathway. Gamma appears to associate with a completed FeMo-co molecule. These results are consistent with a model in which the iron and sulfur from NifB-co associate with NIFNE (where additional steps of FeMo-co biosynthesis likely occur) at some point prior to FeMo-co being associated with gamma and ultimately being incorporated into dinitrogenase. Studies are currently under way to establish where in the biosynthetic pathway molybdenum and homocitrate are incorporated into the FeMo-co.