Nitrogenase Fe protein: A molybdate/homocitrate insertase

Abstract

The Fe protein is indispensable for nitrogenase catalysis and biosynthesis. However, its function in iron-molybdenum cofactor (FeMoco) biosynthesis has not been clearly defined. Here we show that the Fe protein can act as a Mo/homocitrate insertase that mobilizes Mo/homocitrate for the maturation of FeMoco precursor on NifEN. Further, we establish that Mo/homocitrate mobilization by the Fe protein likely involves hydrolysis of MgATP and protein–protein interaction between the Fe protein and NifEN. Our findings not only clarify the role of the Fe protein in FeMoco assembly and assign another function to this multitask enzyme but also provide useful insights into a mechanism of metal trafficking required for the assembly of complex metalloproteins such as nitrogenase.

Nitrogenase catalyzes the nucleotide-dependent reduction of dinitrogen to ammonia, a key step in the global biological nitrogen cycle. The best studied Mo-nitrogenase consists of two components, the Fe protein and the MoFe protein (for recent reviews, see refs. 1–8). The homodimeric Fe protein (encoded by nifH) contains two nucleotide binding sites (one per subunit) and a single [4Fe-4S] cluster at the dimer interface. The α₂β₂-heterotetrameric MoFe protein (encoded by nifD and nifK) contains two sets of unique metal clusters: the P cluster ([8Fe-7S]) (9), which is bridged between each αβ subunit pair, and FeMoco ([Mo-7Fe-9S-X-homocitrate])^¶ (10), which is located within each α-subunit.

The Fe protein plays an essential role in nitrogenase catalysis, serving as the obligate electron donor to the MoFe protein. It is believed that, concomitant with ATP hydrolysis by the Fe protein, electrons are transferred sequentially from the [4Fe-4S] cluster in the Fe protein through the P cluster in the MoFe protein to FeMoco, where substrate reduction takes place. In addition to its catalytic capacity, the Fe protein is also indispensable for the assembly of the complex clusters in the MoFe protein (11). Deletion of the Fe protein-encoding nifH gene results in the formation of a MoFe protein with disrupted P clusters or precursor fragments comprising [4Fe-4S]-like clusters, indicating that the Fe protein may facilitate the fusion of these fragments into a fully assembled [8Fe-7S] P cluster (12–14). Meanwhile, maturation of a NifEN-bound, Mo-free FeMoco precursor into a fully assembled FeMoco requires the presence of Fe protein and MgATP, suggesting that Fe protein may assist FeMoco maturation in a process that involves concurrent MgATP hydrolysis (15). These discoveries are exciting not only because they offer insights into the biosynthetic mechanism of the P cluster and FeMoco but also because they provide evidence that the Fe protein is a multitask component of nitrogenase, participating in catalysis as well as assembly of this important enzyme.

A potential role for the Fe protein in Mo mobilization during FeMoco assembly has been suggested (11, 16, 17); however, the exact function of Fe protein in this process could not be established. Consistent with a role for the Fe protein in Mo sequestration, the first published x-ray structure of the Fe protein of Azotobacter vinelandii contained bound molybdate (18), and it has been reported that radioactively labeled Mo⁹⁹ can accumulate on the Fe protein (19). Recently we developed a “FeMoco maturation assay” in which a Mo-free FeMoco precursor could be converted to a fully assembled FeMoco (15). Based on this assay, we designed a strategy to further investigate the function of Fe protein in FeMoco maturation. Our results show that, in an ATP-dependent process, Fe protein serves as a Mo/homocitrate insertase that delivers and inserts Mo/homocitrate into the Mo-free FeMoco precursor. Our findings not only assign a definitive function to the Fe protein in the assembly of FeMoco but also provide initial insights into the process of heterometal incorporation.

Results and Discussion

Recently, we have shown that NifEN contains a Mo-free FeMoco precursor (15, 17), which can reconstitute and activate the FeMoco-deficient ΔnifB MoFe protein in a so-called “FeMoco maturation assay” (15). Such an assay contains the following: (i) NifEN, the source of FeMoco precursor; (ii) Mo and homocitrate, the missing components from the precursor; (iii) the Fe protein and MgATP, factors facilitating FeMoco maturation in an unknown fashion; and (iv) ΔnifB MoFe protein, the receptor for FeMoco. Based on this assay, we developed a strategy to “uncouple” the original FeMoco maturation assay into several individual steps, allowing further determination of the sequence of events during FeMoco maturation and the roles of particular components in this process. Such a strategy includes the following steps: first, NifEN is incubated with molybdate, homocitrate, Fe protein, and MgATP; then, His-tagged NifEN is repurified from the mixture by affinity chromatography and the nontagged wild-type Fe protein is repurified from the flow-through of the affinity column; and finally, repurified components are tested for their capacities to reconstitute and activate the FeMoco-deficient ΔnifB MoFe protein. Analysis of repurified NifEN (designated NifEN^complete), therefore, enables the determination of the extent of maturation of the NifEN-bound precursor. Based on the study of NifEN^complete, we have established that Mo and homocitrate are incorporated into the precursor while it is bound to NifEN (20). Meanwhile, analysis of the repurified Fe protein from the same incubation mixture (designated Fe protein^complete) allows the determination of whether Fe protein is indeed involved in Mo mobilization during FeMoco biosynthesis and, if so, what type of species it carries. Fe protein^complete was examined by metal and spectroscopic analyses and tested for its capacity as (i) a “Mo/homocitrate donor” in an assay that consists of Fe protein^complete, NifEN, and ΔnifB MoFe protein and (ii) a direct “FeMoco donor” in an assay that consists of only Fe protein^complete and ΔnifB MoFe protein. Control experiments were conducted with Fe proteins repurified from incubation mixtures that lacked one or more of the maturation factors or contained (i) precursor-free ΔnifB NifEN instead of NifEN, (ii) ATP hydrolysis-deficient Fe protein variants instead of wild-type Fe protein, or (iii) ADP, or nonhydrolyzable ATP analogs, instead of ATP. Together with Fe protein^complete, the repurified Fe proteins are categorically designated Fe proteins', with different superscripts indicating different conditions for sample preparation. For designations of Fe proteins', please refer to Materials and Methods.

A Fe protein^complete monomer, like that of the wild-type Fe protein, is ≈30 kDa (data not shown). The molecular mass of Fe protein^complete is ≈60 kDa based on its elution profile on gel filtration Sephacryl S-200 high-resolution column (data not shown), indicating that Fe protein^complete is a homodimer. Fe protein^complete cannot serve as a direct FeMoco donor to the FeMoco-deficient ΔnifB MoFe protein because incubation of Fe protein^complete with ΔnifB MoFe protein alone does not result in the reconstitution and activation of ΔnifB MoFe protein (Table 1). On the other hand, when Fe protein^complete is incubated with NifEN and ΔnifB MoFe protein, ΔnifB MoFe protein is activated to approximately the same extent as it is by a complete FeMoco maturation assay or by incubation with NifEN^complete (Table 1) (15, 20). Further, upon incubation with NifEN and increasing amounts of Fe protein^complete, a maximum activity of ≈300 nmol of C₂H₄ formation per mg of ΔnifB MoFe protein per min is observed (Fig. 1A). These observations indicate that Fe protein^complete is competent in the maturation of the NifEN-bound FeMoco precursor without additional factors. Moreover, given that the precursor does not contain Mo and homocitrate and that no additional Mo and homocitrate are included in the assay, Fe protein^complete appears to be the only source that can provide these two missing components for the transformation of the precursor into a fully assembled FeMoco. This argument is further supported by the observation that Fe protein^{minus Mo/homocitrate}, Fe protein^{minus Mo}, and Fe protein^{minus homocitrate} are not competent in stimulating FeMoco maturation on NifEN (Table 1). These results strongly suggest that Fe protein acts as a Mo/homocitrate insertase that binds and delivers Mo and homocitrate to the Mo-free FeMoco precursor on NifEN.

Table 1.

Reconstitution of MoFe protein with Fe protein'

Assay condition	Activities*
	C2H4 formation under C2H2/Ar	H2 formation under Ar	NH3 formation under N2	H2 formation under N2
FeMoco maturation assay
Complete	290 ± 26 (100)	350 ± 50 (100)	111 ± 20 (100)	65 ± 7 (100)
MoFe protein reconstitution with NifENcomplete†
NifENcomplete	284 ± 17 (98)	375 ± 12 (107)	142 ± 4 (127)	70 ± 7 (108)
MoFe protein reconstitution with Fe protein' as FeMoco donor
Fe proteincomplete	0 (0)	0 (0)	0 (0)	0 (0)
MoFe protein reconstitution with Fe protein' as Mo/homocitrate donor
Fe proteincomplete	281 ± 3 (97)	329 ± 23 (94)	154 ± 6 (137)	66 ± 1 (102)
Fe proteinminus Mo/homocitrate	0 (0)	0 (0)	0 (0)	0 (0)
Fe proteinminus homocitrate	0 (0)	0 (0)	0 (0)	0 (0)
Fe proteinminus Mo	0 (0)	0 (0)	0 (0)	0 (0)
Fe proteinminus MgATP	0 (0)	0 (0)	0 (0)	0 (0)
Fe proteinMgADP	0 (0)	0 (0)	0 (0)	0 (0)
Fe proteinATPγS or Fe proteinAMPPNP	0 (0)	0 (0)	0 (0)	0 (0)
Fe proteinA157S Fe protein‡	16 ± 3 (6)	20 ± 6 (6)	1 ± 0 (<1)	0 (0)
Fe proteinM156C Fe protein‡	0 ± 0 (0)	2 ± 1 (<1)	0 (0)	0 (0)
Fe proteinA157G Fe protein	0 (0)	0 (0)	0 (0)	0 (0)
Fe proteinapo Fe protein	0 (0)	0 (0)	0 (0)	0 (0)
Fe proteinΔnifB NifEN§	37 ± 2 (10)	36 ± 5 (10)	27 ± 2 (24)	5 ± 1 (8)

Data are expressed as nanomoles per minute per milligram of protein. Percentages are given in parentheses.

*The lower detection limits were 0.01, 0.02, 0.001, and 0.02 nmol per min per mg of protein for C₂H₄ formation under C₂H₂/Ar, H₂ formation under Ar, NH₃ formation under N₂, and H₂ formation under N₂, respectively.

^†This assay contained FeMoco-deficient ΔnifB MoFe protein and NifEN^complete. The concentrations of the components and assay conditions are described in a related article (20).

^‡A157S and M156C Fe proteins showed 23% and 16%, respectively, of MgATP hydrolysis activities of wild-type Fe protein (data not shown).

^§Upon incubation with ΔnifB NifEN, Fe protein showed 20% MgATP hydrolysis activity compared with that achieved upon incubation with NifEN (data not shown).

Fig. 1.

Reconstitution assays. (A) MoFe protein reconstitution with Fe protein^complete. Assays were performed as described in Materials and Methods except that the amounts of Fe protein^complete were varied between 0 and 67.5 nmol. Note that in this type of assay, ΔnifB MoFe protein was reconstituted by incubation with NifEN and Fe protein^complete. (B) MoFe protein reconstitution with NifEN^{complete (varied Mo)} (○) and Fe protein^{complete (varied Mo)} (▿). Assays were performed as described in Materials and Methods except that the amounts of molybdate were varied between 0 and 20 μmol. Note that ΔnifB MoFe protein was reconstituted by incubation with NifEN^{complete (varied Mo)} alone (○) or NifEN and Fe protein^{complete (varied Mo)} (▿). The data presented here are the average of three independent experiments. The error bars are indicated.

To establish the stoichiometry of Mo binding to Fe protein, a series of Fe protein^complete and corresponding NifEN^complete samples were prepared as follows. First, 0.5 μmol of NifEN was incubated with 2 μmol of Fe protein, excess MgATP and homocitrate, and various amounts of molybdate ranging from 0 to 20 μmol. Then, Fe protein and NifEN were repurified from the mixture. Finally, the repurified Fe protein samples [designated Fe protein^{complete (varied Mo)}] were assayed by incubation with NifEN and ΔnifB MoFe protein, whereas the repurified NifEN samples [designated NifEN^{complete (varied Mo)}] were assayed by incubation with ΔnifB MoFe protein alone. When the levels of ΔnifB MoFe protein activation by Fe^{complete (varied Mo)} and NifEN^{complete (varied Mo)} are compared, the Mo distribution between the two species in the same incubation mixture can be determined. As shown in Fig. 1B, activation of ΔnifB MoFe protein by NifEN^{complete (varied Mo)} reaches a maximum at ≈1 μmol of molybdate, indicating that NifEN^complete is saturated by Mo at this point. Given that Mo-saturated NifEN^complete contains ≈1 Mo per protein molecule (20), 0.5 μmol of Mo should be assigned to the 0.5 μmol of NifEN^complete in the incubation mixture. Consistent with the association of the remaining 0.5 μmol of Mo with the corresponding Fe protein^complete from the same incubation mixture, ΔnifB MoFe protein is activated when combined with this protein and NifEN (Fig. 1B). Meanwhile, at ≈4.5 μmol of molybdate, activation of ΔnifB MoFe protein by Fe protein^{complete (varied Mo)} reaches approximately the same maximum as that by NifEN^{complete (varied Mo)}, suggesting that both Fe protein^complete and NifEN^complete are saturated by Mo in this case (Fig. 1B). If 0.5 μmol of the ≈4.5 μmol of Mo is assigned to the 0.5 μmol of NifEN^complete in the incubation mixture, then ≈4 μmol Mo can be assigned to the 2 μmol of Fe protein^complete in the same mixture, establishing a stoichiometry of 2 mol Mo per mol Fe protein^complete. This assignment is supported by metal analysis of Mo-saturated Fe protein^complete, which reveals the presence of 1.95 ± 0.02 mol Mo per mol of Fe protein^complete.

Consistent with the presence of Mo or Mo/homocitrate on Fe protein^complete, the EPR spectroscopic features of Fe protein^complete are different from those of the Fe protein in the absence or presence of nucleotides (12). Fe protein^complete exhibits a signal of slightly rhombic line shape in the g ≈ 4 region with g values of 4.44, 4.05, and 3.96 (Fig. 2, spectrum 1), which is absent in the case of Fe protein^{minus Mo/homocitrate} (Fig. 2, spectrum 3) or Fe protein^{minus Mo} (Fig. 2, spectrum 4), indicating a potential association of Mo binding with this signal. It needs to be noted that Fe protein^{minus homocitrate} shows the same signal as Fe protein^complete (Fig. 2, spectrum 2), suggesting that the binding of Mo to Fe protein may not require the presence of homocitrate and that the attachment of homocitrate to Fe protein likely occurs after Mo binding. This observation is consistent with the Mo content of 2.30 ± 0.10 mol of Mo per mol of protein observed for Fe protein^{minus homocitrate}, which is almost identical to that of the Fe protein^complete. A closer examination of the unique EPR signal of Fe protein^complete in the g ≈ 4 region (Fig. 2, spectrum 1, Inset) reveals that it is practically identical in line shape to the EPR signal of NifEN^complete in the same region of the spectrum that likely arises from a minor portion of processed Mo contained within NifEN^complete (20). This observation suggests that a similar processed Mo species is present on the Fe protein, providing further evidence of a function of the Fe protein in delivering Mo and homocitrate to NifEN for FeMoco assembly.

Fig. 2.

EPR spectra of dithionite-reduced Fe proteins'. Spectra of Fe protein^complete (spectrum 1), Fe protein^{minus homocitrate} (spectrum 2), Fe protein^{minus Mo} (spectrum 3), and Fe protein^{minus Mo/homocitrate} (spectrum 4) are shown between 1,000 and 4,000 G. The g values are indicated. The insets of Fe protein^complete (spectrum 1) and Fe protein^{minus Mo} (spectrum 3) show spectra between 1,000 and 2,000 G at a 6-fold magnification. All spectra were measured at a protein concentration of 5 mg/ml as described in Materials and Methods.

Mo K-edge x-ray absorption studies of Fe protein^complete and Fe protein^{minus homocitrate} show that the Mo associated with the Fe protein is modified from the free molybdate (MoO₄²⁻) added to the assay mixture (Fig. 3). The edge position of Fe protein^complete is shifted to lower energy from that of molybdate by ≈2.3 eV (1 eV = 1.602 × 10⁻¹⁹ J), and the intensity of the sharp preedge transition at ≈20,010 eV, which is a characteristic feature of the MoO bonding found in molybdate (21, 22), is diminished (Fig. 3). Proteins involved in prokaryotic Mo homeostasis and transport use a hydrogen-bonding network to bind tetrahedral anionic molybdate (23). X-ray absorption spectroscopy studies of these proteins have shown that the Mo K-edge spectrum of protein-bound molybdate is unchanged from that of free molybdate (24). Thus, the observed differences in the edge spectrum of Fe protein^complete relative to that of molybdate indicate that the Mo associated with the Fe protein is no longer in a MoO₄²⁻ environment (Fig. 3). The spectrum of Fe protein^complete is consistent with a decreased number of MoO bonds (2–3 instead of the 4 found in molybdate) as well as a reduction in the effective oxidation state of Mo due to either a change in the formal oxidation state of Mo or to a change in ligation (25). The spectrum of Fe protein^{minus homocitrate} is very similar to that of Fe protein^complete (Fig. 3), indicating that the interaction of Fe protein with Mo occurs without homocitrate, although the ≈0.5 eV difference between the edges of Fe protein^complete and Fe protein^{minus homocitrate} indicates that homocitrate addition does affect the Mo environment.

Fig. 3.

Normalized Mo K-edge x-ray absorption spectra, with inset smoothed second derivatives, of Fe protein^complete (blue), Fe protein^{minus homocitrate} (red), and molybdate (dotted black). All samples were prepared as described in Materials and Methods.

We established earlier that MgATP hydrolysis is required for FeMoco maturation (15); however, the specific step in FeMoco maturation that involves MgATP hydrolysis was not identified. Our observations that (i) ΔnifB MoFe protein is activated upon incubation with only Fe protein^complete and NifEN (Table 1), (ii) Fe protein^{minus MgATP}, Fe protein^MgADP, Fe protein^ATPγS, and Fe protein^AMPPNP are not able to serve as FeMoco or Mo/homocitrate donors to the precursor on NifEN (Table 1), and (iii) Fe protein^{A157S Fe protein}, Fe protein^{M156C Fe protein}, and Fe protein^{A157G Fe protein} show greatly diminished or no capacity as FeMoco or Mo/homocitrate donors (Table 1) clearly indicate that MgATP hydrolysis is associated with Mo/homocitrate attachment to Fe protein.

Further support for this argument comes from the inorganic phosphate determination, which shows that MgATP is hydrolyzed at a higher rate in a FeMoco maturation assay with molybdate than one without (data not shown), and from Fe chelation experiments (vide infra). The Fe chelation rate constants of Fe protein^{minus Mo/homocitrate}, Fe protein^{minus Mo}, Fe protein^{minus homocitrate}, and Fe protein^complete are 0.008, 0.008, 0.006, and 0.003 s⁻¹ (Fig. 4A, curves 2–5). Given that Fe chelation constants are 0.016 and 0 s⁻¹ for MgATP- and MgADP-bound wild-type Fe protein, respectively (Fig. 4A, curves 1 and 6), the Fe proteins' can be ranked according to their similarity to the MgADP conformation of wild-type Fe protein with the following decreasing order: Fe protein^complete > Fe protein^{minus homocitrate} > Fe protein^{minus Mo} ≈ Fe protein^{minus Mo/homocitrate}. This observation yet again suggests a higher turnover rate of MgATP by Fe protein in the presence of Mo and an association between MgATP hydrolysis and Mo binding to the Fe protein.

Fig. 4.

Fe chelation of the Fe protein [4Fe-4S]¹⁺ cluster. (A) The formation of the complex between the Fe chelator bathophenanthroline disulfonate and Fe from the [4Fe-4S]¹⁺ cluster of the Fe protein was measured at 535 nm for the following samples: Fe protein in the presence of MgATP (curve 1), Fe protein^{minus Mo/homocitrate} alone (curve 2), Fe protein^{minus Mo} alone (curve 3), Fe protein^{minus homocitrate} alone (curve 4), Fe protein^complete alone (curve 5), and Fe protein in the presence of MgADP (curve 6). All protein concentrations were 0.8 mg/ml. Final concentrations of 0.2 mM ATP or ADP and 0.4 mM MgCl₂ were used for sample 1 and sample 6. Note that the levels of chelation are almost identical for Fe protein^{minus Mo} (curve 2) and Fe protein^{minus Mo/homocitrate} (curve 3). (B) Protection of Fe protein from Fe chelation by MoFe protein and NifEN. Fe chelation of Fe protein in the presence of MgATP alone (curve 1), MgATP and NifEN (curve 2), MgATP and MoFe protein (curve 3), and MgADP alone (curve 4) are shown. The concentrations of NifEN and MoFe protein were 0.13 mg/ml. All other concentrations are the same as described in A. Curves were fitted to single exponential equations over a period of 20 s, resulting in the following rate constants: 0.016 s⁻¹(A, curve 1), 0.008 s⁻¹ (A, curve 2), 0.008 s⁻¹ (A, curve 3), 0.006 s⁻¹ (A, curve 4), 0.003 s⁻¹ (A, curve 5), 0 s⁻¹ (A, curve 6), 0.016 s⁻¹ (B, curve 1), 0.009 s⁻¹ (B, curve 2), 0.002 s⁻¹ (B, curve 3), and 0 s⁻¹ (B, curve 4).

Having established the requirement of MgATP hydrolysis for Mo binding to the Fe protein, the next question to address is whether there is electron transfer concurrent with MgATP hydrolysis in this process. As shown in Table 1, Fe protein^{apo Fe protein} is unable to act as a Mo/homocitrate donor, indicating that the [4Fe-4S] cluster is indispensable for Mo/homocitrate attachment to Fe protein. This result is also consistent with our earlier observation that the reductant dithionite is absolutely required along with the Fe protein and MgATP for FeMoco maturation (15). Interestingly, Fe protein^{ΔnifB NifEN} is able to “load” a certain amount of Mo and homocitrate that can be subsequently inserted into the precursor on NifEN, as evidenced by activation of the ΔnifB MoFe protein by Fe protein^{ΔnifB NifEN}, which is 24% of that activated by Fe protein^complete. Consistent with the FeMoco maturation results with Fe protein^{ΔnifB NifEN}, the level of MgATP hydrolysis by wild-type Fe protein incubated with ΔnifB NifEN is ≈20% of that achieved by incubation with NifEN (data not shown). Considering that NifEN is structurally analogous to the MoFe protein (26), it is plausible that, in a process analogous to Fe protein-MoFe protein docking during catalysis, Mo is mobilized through a MgATP-induced Fe protein–NifEN interaction associated with electron transfer from the [4Fe-4S] cluster of the Fe protein to the permanent clusters of NifEN and finally to the FeMoco precursor.

The interaction between NifEN and the Fe protein is observed in chelation experiments (Fig. 4B), which show that the “exposed” [4Fe-4S]¹⁺ cluster of the Fe protein in the MgATP conformation (Fig. 4B, curve 1) is “protected” by the presence of NifEN (Fig. 4B, curve 2), resulting in a significant decrease of in the chelation rate constant from 0.016 to 0.009 s⁻¹. A similar effect, albeit to a more dramatic extent, is observed when Fe protein is incubated with the MoFe protein and MgATP (Fig. 4B, curve 3), in which case a chelation rate constant of 0.002 s⁻¹ is measured. The different levels of protection provided by NifEN and MoFe protein suggest that the Fe protein interacts with these two proteins differently, which is not surprising given that NifEN is not a perfect MoFe protein homolog (26) and that different reactions are taking place during these interactions.

The interaction between the protein mobilizing the heterometal and the particular scaffold protein carrying the precursor may also direct the insertion of the appropriate heterometal into the cofactors of other nitrogenases.^¶ For example, the incorporation of V into the FeVco of V-nitrogenase likely requires the Fe protein homologue, VnfH, and the NifEN homologue, VnfEN; whereas in the case of the FeFeco of Fe-only nitrogenase, in which no heterometal insertion is necessary, no specific scaffold protein has been identified (11). Additionally, nucleotide-dependent metal insertion is well documented for the molybdopterin-containing enzymes, certain carbon monoxide dehydrogenases, and some hydrogenases (28–30). Thus, the Fe protein-mediated Mo mobilization bears implications for the general mechanism of heterometal incorporation.

In summary, we have established that the Fe protein can act as a Mo/homocitrate insertase that mobilizes Mo/homocitrate for the maturation of FeMoco precursor on NifEN. Further, we have shown that Mo/homocitrate mobilization by the Fe protein likely involves hydrolysis of MgATP and protein–protein interaction between the Fe protein and NifEN (see Fig. 5, which is published as supporting information on the PNAS web site). Our findings not only clarify the role of the Fe protein in FeMoco assembly and assign another function to this multitask enzyme, but also provide useful insights into a mechanism of metal trafficking required for the assembly of complex metalloproteins like nitrogenase. Future investigations involve studies of focus on the nature of the protein–protein interactions between the Fe protein and the NifEN complex, which will hopefully lead to the elucidation of a more definitive mechanism of Mo/homocitrate insertion by the Fe protein.

Materials and Methods

Unless otherwise noted, all chemicals and reagents were obtained from Fisher, Aldrich, or Sigma.

Cell Growth and Protein Purification.

All A. vinelandii strains were grown in 180-liter batches in a 200-liter New Brunswick Scientific fermentor on Burke's minimal medium supplemented with 2 mM ammonium acetate. The growth rate was measured by cell density at 436 nm by using a 20 Genesys Spectrophotometer (Spectronic, Westbury, NY). After ammonia consumption, the cells were derepressed for 3 h followed by harvesting by using a flow-through centrifugal harvester (Cepa, Lahr/Schwarzwald, Germany). The cell paste was washed with 50 mM Tris·HCl (pH 8.0). Published methods were used for the purification of all Fe proteins (31), wild-type MoFe protein (32), His-tagged ΔnifB MoFe protein (12), His-tagged NifEN, and His-tagged ΔnifB NifEN (15).

EPR Spectroscopy.

All EPR samples were prepared in a Vacuum Atmospheres (Hawthorne, CA) dry box with an oxygen level of <4 ppm. All dithionite-reduced samples were in 25 mM Tris·HCl (pH 8.0), 10% glycerol, and 2 mM Na₂S₂O₄. Indigo disulfonate -oxidized samples were prepared as described (31). Samples were either used as they were or they were concentrated in a Centricon-30 (Amicon) in anaerobic centrifuge tubes outside of the dry box. All EPR spectra were recorded by using a ESP 300 E_z spectrophotometer (Bruker, Billerica, MA) and interfaced with an ESR-9002 liquid helium continuous flow cryostat (Oxford Instruments, Oxon, U.K.). All spectra were recorded at 13 K by using a microwave power of 50 mW, a gain of 5 × 10⁴, a modulation frequency of 100 kHz, and a modulation amplitude of 5 G. A microwave frequency of 9.43 GHz was used to record 10 scans for each sample.

X-Ray Absorption Spectroscopy.

All Fe protein x-ray absorption spectroscopy samples were prepared in a Vacuum Atmospheres dry box with an oxygen level of <4 ppm. Samples of Fe protein^complete (5.2 mg/ml, ≈0.2 mM Mo) and Fe protein^{minus homocitrate} (4.6 mg/ml, ≈0.2 mM Mo) in 25 mM Tris·HCl (pH 8.0), 500 mM NaCl, 2 mM Na₂S₂O₄, and 33% glycerol were loaded into 250-μl, 15-mm pathlength nylon cells and flash-frozen in a pentane/liquid N₂ slush. Sodium molybdate, 8 mM in 33% glycerol/water, was measured in a similar nylon cell. X-ray absorption spectroscopy data were measured at the Stanford Synchrotron Radiation Laboratory under 3 GeV, 80–100 mA beam conditions by using focused beamline 9-3 with a Si (220) double-crystal monochromator. The samples were maintained at 10 K during data collection in an Oxford Instruments CF1208 continuous-flow liquid He cryostat. Data were collected as Mo Kα fluorescence by using a Canberra 30-element solid-state Ge detector. Radiation from elastic/inelastic scattering and Kβ fluorescence were minimized at the detector by Soller slits with a Zr filter between the cryostat and the detector. The x-ray energy was calibrated to the inflection point at 20,003.9 eV of a standard Mo foil measured concurrent with the samples. Changes in the edge position over time, which would indicate photoreduction of an oxidized metal site, were not observed for any sample. Using the program xfit (33), a background polynomial absorption curve was subtracted from data over the range 19,690–20,410 eV, and the resulting spectra were normalized to have an edge-jump of 1.0 at 20,025 eV.

FeMoco Maturation Assay.

Assays designed to convert the NifEN-bound FeMoco precursor to mature FeMoco contained, in a 0.8-ml total volume, 25 mM Tris·HCl (pH 8.0), 20 mM Na₂S₂O₄, 0.5 mg of purified FeMoco-deficient ΔnifB MoFe protein from A. vinelandii strain DJ1143 (34), 1.4 mg of Fe protein, 0.3 mM homocitrate, 0.3 mM sodium molybdate, 0.8 mM ATP, 1.6 mM MgCl₂, 10 mM creatine phosphate, and 8 units of creatine phosphokinase. FeMoco maturation was initiated with the addition of 2 mg of isolated FeMoco precursor-containing NifEN (15) to the mixture mentioned above. Such a reaction mixture was incubated at 30°C for 30 min, stopped by the addition of 40 nmol (NH₄)₂MoS₄ (15), and determined for enzymatic activities as described (32, 35, 36). The reaction products H₂ and C₂H₄ were analyzed as published elsewhere (32), whereas ammonium was determined by using a high performance liquid chromatography fluorescence method (37). Homocitrate lactone (Sigma) containing an mixture of the stereochemical configurations R (≈95%) and S (≈5%) was converted to the free acid as described elsewhere (15).

Preparation of Fe Protein'.

As described elsewhere (20), maturation of FeMoco on NifEN was monitored by incubating FeMoco precursor-bound NifEN with modified FeMoco maturation assays, repurifying His-tagged NifEN by affinity chromatography (categorically designated NifEN′), and determining the activities of NifEN′ subsequently. Here, the Fe proteins of the same incubation assays were repurified from the flow-through of the affinity columns and analyzed for activities (below). Such repurified Fe proteins are categorically designated Fe protein', and different superscripts are used to indicate different maturation assay compositions as follows. (i) Fe protein^complete, the assay contained, in a 50-ml total volume, 25 mM Tris·HCl (pH 8.0), 2 mM Na₂S₂O₄, 100 mg of FeMoco precursor containing NifEN, 120 mg of wild-type Fe protein, 0.4 mM homocitrate, 0.4 mM sodium molybdate, 2.4 mM ATP, 4.8 mM MgCl₂, 30 mM creatine phosphate, and 24 units/ml of creatine phosphokinase. This mixture was stirred for 1 h at 30°C, and then the Fe protein was repurified as described (15). Note that Fe protein, MgATP, molybdate, and homocitrate were in excess to NifEN in terms of molar ratios so that the Fe protein was still “loaded” with other components of the FeMoco maturation assay. (ii) Fe protein^{ΔnifB NifEN} assay was a control assay with the same conditions as i except that NifEN was replaced by precursor-free ΔnifB NifEN (15). (iii) Fe protein^{minus Mo/homocitrate}, Fe protein^{minus homocitrate}, Fe protein^{minus Mo}, and Fe protein^{minus MgATP} assays were control assays with the same conditions as i except that one or more of the components required for FeMoco maturation were omitted. (iv) Fe protein^{apo Fe protein}, Fe protein^{A157 Fe protein}, Fe protein^{M156C Fe protein}, and Fe protein^{A157G Fe protein} assays were control assays with the same conditions as i except that the wild-type Fe protein was replaced by 120 mg of apo, A157S, M156C, or A157G Fe proteins, respectively. (v) Fe protein^MgADP, Fe protein^ATPγS, and Fe protein^AMPPNP assays were control assays with the same conditions as i except that ATP was replaced by 2.4 mM ADP, ATPγS [adenosine 5′-O-(3-thiotriphosphate)] or AMPPNP (5′-adenylylimido-diphosphate). Creatine phosphate and creatine phosphokinase were omitted when the function of ADP was evaluated. Table 2, which is published as supporting information on the PNAS web site, summarizes the designations of repurified Fe protein' and the assay conditions used to obtain these proteins.

MoFe Protein Reconstitution Assay with Fe Protein'.

(i) The assay designed to test whether Fe protein' can serve as “Mo/homocitrate donor” to the NifEN-bound precursor contained, in a 0.8-ml total volume, 25 mM Tris·HCl (pH 8.0), 20 mM Na₂S₂O₄, 2.5 mg of isolated Fe protein', and 0.5 mg of purified ΔnifB MoFe protein from A. vinelandii strain DJ1143 (34). FeMoco insertion was initiated with the addition of 2 mg of isolated, FeMoco precursor containing NifEN (15) to the mixture mentioned above. Such a reaction mixture was incubated, stopped, and then analyzed for enzymatic activities as described for FeMoco maturation assay (above). (ii) The assay designed to test whether Fe protein' can serve as direct “FeMoco donor” to the ΔnifB MoFe protein contained, in a 0.8-ml total volume, 25 mM Tris·HCl (pH 8.0), 20 mM Na₂S₂O₄, and 0.5 mg purified ΔnifB MoFe protein from A. vinelandii strain DJ1143 (34). FeMoco insertion was initiated with the addition of 2.5 mg of isolated Fe protein' to the mixture above. Such a reaction mixture was incubated, stopped, and then determined for enzymatic activities as described above for FeMoco maturation assay.

Metal Analysis, Fe-Chelating Assay, and Preparation of Apo Fe Protein.

Mo (38) and Fe (39) contents were determined as published elsewhere. Potential protein–protein interactions between NifEN and the Fe protein were examined by Fe-chelating assays as described elsewhere (40). Apo Fe protein was prepared by chelating the [4Fe-4S] cluster from the Fe protein with the Fe chelator bathophenanthroline disulfonate. The chelation reaction contains, in a 10-ml total volume, 25 mM Tris·HCl (pH 8.0), 20 mM ATP, 40 mM MgCl₂, 20 mM bathophenanthroline disulfonate, and 100 mg of wild-type Fe protein. The reaction mixture was incubated for 15 min, and the Fe protein was subsequently passed over a 2.5 × 100-cm Ultrogel AcA34 (Ciphergen, Fremont, CA) gel filtration column in 25 mM Tris·HCl (pH 8.0). The resulting Fe protein was completely Fe-deficient (or apo) based on Fe analysis.

Abbreviations

FeMoco

iron-molybdenum cofactor

ΔnifB NifEN

tetrameric NifEN complex produced by A. vinelandii nifB-deletion strain YM9A

ΔnifB MoFe protein

MoFe protein produced by A. vinelandii nifB-deletion strain DJ1143