A Proposed Role for the Azotobacter vinelandii NfuA Protein as an Intermediate Iron-Sulfur Cluster Carrier*

Iron-sulfur clusters ([Fe-S] clusters) are assembled on molecular scaffolds and subsequently used for maturation of proteins that require [Fe-S] clusters for their functions. Previous studies have shown that Azotobacter vinelandii produces at least two [Fe-S] cluster assembly scaffolds: NifU, required for the maturation of nitrogenase, and IscU, required for the general maturation of other [Fe-S] proteins. A. vinelandii also encodes a protein designated NfuA, which shares amino acid sequence similarity with the C-terminal region of NifU. The activity of aconitase, a [4Fe-4S] cluster-containing enzyme, is markedly diminished in a strain containing an inactivated nfuA gene. This inactivation also results in a null-growth phenotype when the strain is cultivated under elevated oxygen concentrations. NifU has a limited ability to serve the function of NfuA, as its expression at high levels corrects the defect of the nfuA-disrupted strain. Spectroscopic and analytical studies indicate that one [4Fe-4S] cluster can be assembled in vitro within a dimeric form of NfuA. The resultant [4Fe-4S] cluster-loaded form of NfuA is competent for rapid in vitro activation of apo-aconitase. Based on these results a model is proposed where NfuA could represent a class of intermediate [Fe-S] cluster carriers involved in [Fe-S] protein maturation.

The in vivo maturation of simple [Fe-S] proteins is proposed to require preassembly of [Fe-S] species on molecular scaffolds. The first [Fe-S] cluster assembly system to be described is the NIF 2 system from Azotobacter vinelandii. This system consists of a cysteine desulfurase, encoded by nifS, which supplies the S for [Fe-S] cluster formation, and a proposed scaffold protein, encoded by nifU (1). The NIF system is specialized for the maturation of [Fe-S] proteins involved in nitrogen fixation.
A. vinelandii also contains a second [Fe-S] protein maturation system designated ISC. The ISC system is required for the general maturation of cellular [Fe-S] proteins involved in intermediary metabolism, such as aconitase (2). The ISC system is more complicated than the NIF system as it includes the products of eight contiguous genes: iscR, iscS, iscU, iscA, hscB, hscA, fdx, and iscX (3). Although the NIF and ISC systems exhibit physiological target specificity, each can partially replace the function of the other, when expressed at high levels (4,5).
Even though the NIF and ISC systems are differentiated by their apparent target specificities, they share a number of common structural and functional features. For example, NifS and IscS have similar sequences, and they both exhibit cysteine desulfurase activity (2). IscU also shares considerable sequence identity when compared with the N-terminal domain of NifU, including conservation of three cysteine residues that are likely to provide the nucleation site(s) for [Fe-S] cluster assembly (2,6).
NifU is a modular protein that contains three distinct domains (Fig. 1). The central domain contains a stable redoxactive [2Fe-2S] cluster with an as-yet-unknown function (7). In vitro and in vivo experiments have established that labile [Fe-S] clusters can be assembled on both the N-terminal and C-terminal domains of NifU, and such cluster-loaded forms of NifU can be used for activation of the nitrogenase Fe-protein (8,9). Thus, NifU contains two different sites upon which labile [Fe-S] clusters can be assembled in vitro, but the functional relationship between these sites is not yet known.
There are no genes within the ISC transcriptional unit that encode proteins with sequence similarity to the C-terminal domain of NifU. However, located elsewhere on the A. vinelandii genome is a gene, designated nfuA, whose product encodes a protein having a C-terminal sequence similar to the C-terminal domain of NifU (Fig. 1). The sequence conservation between NifU and NfuA includes two cysteine residues that are required for the in vitro assembly of [Fe-S] clusters within the NifU C-terminal domain. Like NifU, NfuA also appears to be a modular protein because the amino acid sequence within its N-terminal region shares some sequence similarity with another protein involved in [Fe-S] protein maturation designated IscA (Fig. 1). IscA is a nonessential protein encoded within the ISC transcriptional unit, and it has been proposed to serve as an alternative [Fe-S] cluster assembly scaffold (10 -12) or as an iron donor during the [Fe-S] cluster assembly process (13). IscA contains three cysteine residues that are essential for its physiological function (3). However, sequence conservation between IscA and the N-terminal domain of NfuA does not include the three cysteine residues contained within IscA.
Sequence conservation between the N-terminal region of NfuA and the IscA family of proteins, as well as the conservation between the C-terminal domains of NfuA and NifU suggest that NfuA could play a role in [Fe-S] protein maturation. This possibility is supported by evidence that [Fe-S] clusters can be assembled on NfuA-like proteins from other organisms (14 -17). In some cases, loss of NfuA-like function impairs [Fe-S] protein maturation (17)(18)(19)(20). In the present work we have performed a genetic, biochemical, and spectroscopic analysis of NfuA from A. vinelandii to gain insight into its cellular function in this organism.

EXPERIMENTAL PROCEDURES
Materials-Materials used in this work were purchased from Sigma, Fisher Scientific, New England Biolabs, Invitrogen, or from other sources as indicated throughout the text.
Bioinformatics and DNA Analysis-Sequence alignments were performed using the SDSC Biology Workbench program. The Lasergene 7 software was used to analyze DNA and protein sequences. When necessary, A. vinelandii genomic DNA was extracted, amplified, and sequenced as described previously (5).
Plasmid and Strain Construction-Construction of A. vinelandii strains via transformation with appropriate plasmids has been described previously (21). In this work, DJ1707 was obtained by transformation of our wild-type laboratory strain designated DJ (ATCC BAA-1303) with pDB1598, a plasmid that includes a kanamycin resistance cartridge inserted into the XhoI sites of nfuA. This construct results in deletion of a 290-bp fragment from the nfuA gene. DJ1759 and DJ1769, respectively, produce the NfuA C152A -and NfuA C155A -substituted forms of NfuA and were constructed using plasmids pDB1610 and pDB1614, which contain site-directed cysteine to alanine substitutions (codon change of UGC to GCC) at residue positions 152 and 155, respectively. Site-directed mutagenesis procedures were conducted using the GeneEditor TM system by Pro-mega. The parent plasmid used for deletion, insertion, or amino acid substitutions was pDB1577, which contains the intact nfuA gene, as well as 570-bp downstream of the gene, in the pUC-7 vector.
Abundant production and controlled expression of NifUS and their variants were achieved by placing the desired genes under the control of the inducible arabinose regulatory elements as described previously (5). Plasmid pDB1598 was used to inactivate nfuA in DJ1626, a strain where nifUS is placed under the control of the araBAD promoter with concomitant in-frame deletion of the nif-regulated nifUS (5). This inactivation yielded strain DJ1772. Similarly, strains DJ1773 and DJ1791 were constructed, which contain nifU variants with cysteine to alanine substitutions at residues 35 and 275, respectively, and are otherwise isogenic to DJ1772.
Plasmid pDB1417 was designed by insertion of nfuA, flanked by NdeI and BamHI restriction sites, into the NdeI and BamHI sites of pT7-7 for recombinant expression. Similarly, plasmids pDB1582 and pDB1583 were constructed for recombinant expression of NfuA 152CA and NfuA 155CA , respectively.
Cell Growth at Ambient and Elevated Oxygen Levels-All strains used in this study were cultured in Burk's medium supplemented with 13 mM ammonium acetate (22). When appropriate, media were also supplemented with 20 mM L-arabinose. To expose cells to elevated oxygen concentrations, Petri plates were placed in BBL GasPak jars and filled with a gas mixture containing 40% O 2 as described previously (3).
Aconitase and Isocitrate Dehydrogenase Assays of A. vinelandii Crude Extracts-A. vinelandii cells were grown in liquid cultures containing Burk's medium supplemented with 13 mM ammonium acetate until they reached A 600 ϭ 0.5. Cells were subsequently harvested by centrifugation at 5000 rpm for 5 min, and stored at Ϫ20°C, if not used immediately. To prepare crude extracts, cells were first resuspended in degassed buffer containing 25 mM Tris at pH 7.4 and were lysed using a french pressure cell at 12,000 psi. Lysed cells, which were maintained in airtight vials filled with Argon gas, were centrifuged at 35,000 rpm for 45 min. Crude extracts were separated from the cell debris pellet inside a Coy anaerobic glove box containing 4% H 2 gas and 96% N 2 to minimize exposure to oxygen. Aconitase assays were prepared in 1-ml quartz cuvettes (sealed airtight with rubber septa) containing 700 l of 100 mM Tris-HCl buffer at pH 8.0 and 0.1 mg of crude extract. The reaction was initiated with the addition of 25 mM sodium citrate, and activity was monitored at 240 nm (23). Isocitrate dehydrogenase assays of crude extracts were performed in quartz cuvettes containing 900 l of 50 mM MOPS buffer at pH 7.3, 2.5 mM MgCl 2 , 0.45 mM NADP ϩ , and 0.1 mg of crude extract. The reaction was started with the addition of 80 mM DL isocitric acid (trisodium salt) in the MOPS/MgCl 2 buffer (described above) and monitored at 340 nm (24).
Purification of NfuA and Its Variants-Escherichia coli BL21(DE3) cells transformed with plasmids pDB1417, pDB1582, or pDB1583 were used for the heterologous expression of NfuA, NfuA 152CA , or NfuA 155CA , respectively. Cells were grown in LB medium containing 0.1 mg/ml ampicillin until they reached A 600 ϭ 0.5, at which point, they were induced with 29 mM lactose and allowed to continue to grow for 3 h at 30°C. Cells were then harvested by centrifugation at 5000 rpm for 5 min and stored at Ϫ20°C until use. Cells expressing recombinant NfuA or its alanine-substituted variants were resuspended in degassed 50 mM Tris-HCl, pH 7.4 buffer (buffer A) and lysed either by sonication (1-min pulse/30-s pause for a total of 20 min) or by a french pressure cell. Crude extracts were prepared as described above. All purification procedures were conducted anaerobically with Schlenk lines filled with argon gas. Crude extracts were loaded at 2.5 ml/min using a peristaltic pump with a 25-ml Q-Sepharose column previously equilibrated with 5 bed volumes of buffer A. The column was subsequently washed with 50 ml of buffer A and 75 ml of buffer A containing 100 mM NaCl. To elute NfuA, 200 ml of 0.1-1.0 M NaCl gradient were applied, and the contents of each fraction were analyzed by 15% SDS-PAGE. The fractions containing NfuA, which eluted between 40 and 94 ml of the gradient, were consolidated, and (NH 4 ) 2 SO 4 was added to a final concentration of 1 M. This sample was then loaded onto a 25-ml phenyl-Sepharose column that was previously equilibrated with buffer A containing 1 M (NH 4 ) 2 SO 4 . Following binding of the sample onto the column, the latter was washed with 125 ml of buffer A containing 1 M (NH 4 ) 2 SO 4 followed by 25 ml of 0.5 M (NH 4 ) 2 SO 4 in buffer A. The protein was then eluted using a 100-ml gradient of 0.5 to 0.0 M (NH 4 ) 2 SO 4 followed by 50 ml of buffer A. The fractions containing NfuA, collected from the last 28 ml of the gradient, were then concentrated by Amicon ultrafiltration using a YM10 membrane. The concentrated sample was loaded onto a 320-ml Superdex S75 gel filtration column equilibrated with 100 mM Tris-HCl buffer (pH 8.0) with 1 mM dithiothreitol (buffer B) and 150 mM NaCl. The purest fractions of NfuA, as judged by SDS-PAGE, were pooled and concentrated by Amicon ultrafiltration using a YM10 membrane, and frozen as pellets in liquid N 2 until used. The resulting samples of NfuA were Ͼ95% pure as judged by gel densitometry.
In Vitro Reconstitution of NfuA-Purified NfuA, NfuA 152CA , or NfuA 155CA (0.24 mM) were incubated with NifS (12 M), ferrous ammonium sulfate (4.8 mM), and L-cysteine (4.8 mM) in buffer B for 3 h at room temperature under argon inside a Vacuum Atmospheres glove box (Ͻ2 ppm O 2 ). Excess reagents were removed from the sample by loading it onto a Hi-Trap Q-Sepharose column and eluting with a 0 to 1.0 M NaCl gradient using buffer B. The holoprotein eluted between 0.45 and 0.55 M NaCl and was concentrated by ultrafiltration using an Amicon YM10 membrane. 57 Fe-enriched ferrous sulfate (Ͼ95% enrichment) was used in place of natural abundance ferrous ammonium sulfate in the preparation of Mössbauer samples.
Determination of the Oligomeric State of Apo-and Holo-NfuA-The oligomeric states of both apo-and holo-NfuA were determined by gel-filtration chromatography using a 25-ml Superdex S75 column. The elution buffer was buffer B with 100 mM KCl, which was applied to the column at a flow rate of 0.4 ml/min. Molecular weight standards used were blue dextran (M r 2,000,000), ␤-amylase (M r 200,000), alcohol dehydrogenase (M r 150,000), bovine serum albumin (M r 66,000), carbonic anhydrase (M r 29,000), and cytochrome c (M r 12,400).
Analytical and Spectroscopic Analyses-Protein concentrations were determined by the DC protein assay (Bio-Rad) using bovine serum albumin as the standard. Iron concentrations were determined colorimetrically using bathophenanthroline under reducing conditions, after digestion of the protein in 0.8% KMnO 4 /0.2 M HCl (25) or by using the commercial Quantichrom TM iron assay kit (DIFE-250) from Bioassay Systems. Samples for spectroscopic studies were prepared and handled under argon in a Vacuum Atmospheres glove box (O 2 Ͻ 2 ppm). UV-visible absorption spectra were recorded at room temperature using a Shimadzu UV-3101PC or Cary 50 Bio spectrophotometer. Resonance Raman spectra were recorded as previously described, using an Instrument SA Raman or U1000 spectrometer coupled with a Coherent Sabre argon ion laser, with 20-l frozen droplets of 2-3 mM sample mounted on the cold finger of an Air Products Displex Model CSA-202E closed cycle refrigerator (26). Signal-to-noise was improved by multiple scans and bands due to the frozen buffer solution were subtracted from all the spectra shown in this work after normalization of lattice modes of ice centered at 230 cm Ϫ1 . Variable temperature magnetic circular dichroism (VTMCD) spectra were recorded using samples containing 55% (v/v) ethylene glycol in 1-mm cuvettes using an Oxford Instruments Spectromag 4000 (0 -7 T) split-coil superconducting magnet (1.5-300 K) mated to a Jasco J-715 spectropolarimeter (27). X-band (ϳ9.6 GHz) EPR spectra were recorded using a Bruker ESP-300E EPR spectrometer equipped with a dual-mode ER-4116 cavity and an Oxford Instruments ESR-9 flow cryostat. Mössbauer spectra were recorded using the previously described instrumentation (28). Analysis of the Mössbauer data was performed with the program WMOSS (Web Research).
Activation of Apo-aconitase by Reconstituted NfuA-Apo-aconitase was prepared by incubating recombinantly expressed and purified A. vinelandii aconitase A (AcnA) (29) with EDTA and potassium ferricyanide as described previously (30). Activation mixtures contained 4 M apo-AcnA and between 0 and 12 M [4Fe-4S] cluster-loaded NfuA in buffer B. The concentration of [4Fe-4S] cluster-loaded NfuA corresponds to the concentration of [4Fe-4S] 2ϩ clusters calculated by using a molar extinction coefficient of ⑀ 400 ϭ 15.0 mM Ϫ1 cm Ϫ1 per [4Fe-4S] 2ϩ cluster. Activation mixtures were incubated at room temperature (22°C) under anoxic conditions, and 10-l samples were taken at different time points and assayed for AcnA activity. AcnA activity was measured spectrophotometrically at 240 nm at 22°C by following the formation of cis-aconitate from citrate or isocitrate, using a molar absorption coefficient ⑀ 240 of 3400 mM Ϫ1 cm Ϫ1 for cis-aconitate (23). Assays (1 ml) were carried out in sealed anoxic cuvettes containing 900 l of 100 mM Tris/HCl (pH 8.0) and initiated by addition of 100 l of 200 mM citrate or isocitrate. Anaerobically reconstituted samples of A. vinelandii AcnA containing one [4Fe-4S] cluster per protein monomer exhibited maximal specific activity of 25 units/mg using citrate (ϳ100% activity) and 79 units/mg using isocitrate (ϳ100% activity) (29). The time course of holo-AcnA formation at 22°C was analyzed by fitting to second order kinetics, based on the initial concentrations of apo-AcnA and [4Fe-4S] 2ϩ clusters on NfuA, using the Chemical Kinetics Simulator software package (IBM).

In Vitro Assembly of a [4Fe-4S] Cluster within NfuA-When
A. vinelandii NfuA is heterologously produced in E. coli and purified under anoxic conditions, it does not contain a [Fe-S] cluster. Gel exclusion chromatography indicated that as-isolated, recombinant NfuA is an approximately equal mixture of dimeric and tetrameric species (data not shown). When as-isolated NfuA is incubated with NifS, L-cysteine, and Fe 2ϩ , a [Fe-S] cluster is assembled. Gel exclusion chromatography revealed that the cluster-bound form of NfuA is resolved into a single dimeric species. Analytical data, coupled with the UV-visible absorption spectrum and extinction coefficients (Fig. 2), indicate that the reconstituted sample contains approximately one While the absorption and Mössbauer results provide unambiguous evidence for anaerobic reconstitution of a [4Fe-4S] 2ϩ cluster on NfuA, the vibrational properties, as determined by resonance Raman spectroscopy, are somewhat atypical compared with well-characterized biological [4Fe-4S] 2ϩ centers (31)(32)(33) and exhibit characteristics that have been observed previously only for the subunit-bridging [4Fe-4S] 2ϩ cluster in the MgATP-bound nitrogenase iron protein (34). This feature is illustrated in Fig. 4, which shows a comparison of the resonance Raman spectra of the [4Fe-4S] 2ϩ clusters in reconstituted forms of A. vinelandii IscU and NfuA in the Fe-S stretching region using 457-nm excitation. The major difference lies in the anomalously high frequency for the most intense band in the spectrum (353 cm Ϫ1 for the NfuA [4Fe-4S] 2ϩ center compared with 335-343 cm Ϫ1 in other biological [4Fe-4S] 2ϩ centers), which is generally assigned to the symmetric breathing mode of the [4Fe-4S] cubane core. Analogous spectra with similar band frequencies and relative intensities were observed for the reconstituted NfuA using 488-and 514-nm excitation, indicating that this behavior cannot be attributed to anomalous excitation profiles for discreet bands. One possible explanation is that the intense bands at 353 and 359 cm Ϫ1 for the [4Fe-4S] 2ϩ center in NfuA correspond to asymmetric Fe-S(Cys) stretching modes and that the symmetric breathing modes primarily involving the [4Fe-4S] core and the Fe-S(Cys) bonds are only weakly enhanced as a result of distortions of the core and/or the cluster environment. This possibility is tentatively supported by the observation of similar frozen solution resonance Raman spectra for the [4Fe-4S] 2ϩ cluster in MgATP-bound nitrogenase iron protein, which was found to be split into two [2Fe-2S] fragments separated by ϳ5 Å in the crystalline state (34). Normal mode calculations coupled with 57/54 Fe and 34/32 S isotope  shifts and measurement of depolarization ratios are planned to test this hypothesis and to provide vibrational assignments for the [4Fe-4S] 2ϩ center in NfuA.
The nature of the cluster ligation for the [4Fe-4S] center on NfuA was addressed by amino acid substitution of the two conserved cysteine residues located within the Cys-X-X-Cys motif. It was not possible to reconstitute a [Fe-S] cluster on variant forms of NfuA proteins that have either of these two cysteine residues substituted by alanine (data not shown). This information, together with the analytical and spectroscopic data, are reasonably interpreted in terms of cluster-loaded NfuA having a [4Fe-4S] 2ϩ cluster that is symmetrically bridged between two identical subunits and coordinated by conserved cysteine residues from opposing subunits.
The redox properties of the [4Fe-4S] 2ϩ cluster assembled on NfuA were assessed using a combination of UV-visible absorption, EPR, and VTMCD spectroscopy (Figs. 2 and 5). Anaerobic addition of one reducing equivalent of dithionite results in reversible bleaching of the visible absorption (Fig. 2) 2). The low spin quantification is in accord with the partial bleaching of the visible absorption and appears to reflect partial reduction. No low-field EPR signals indicative of a S ϭ 3/2 [4Fe-4S] ϩ clusters were observed. In accord with the EPR results, the VTMCD spectrum is consistent with a S ϭ 1/2 [4Fe-4S] ϩ cluster rather than a S ϭ 3/2 [4Fe-4S] ϩ cluster (35), and the intensity indicates ϳ0.4 [4Fe-4S] ϩ clusters/NfuA dimer. Partial reduction by one reducing equivalent of dithionite implies a redox potential for the [4Fe-4S] 2ϩ,ϩ couple that is close to that of dithionite at pH 8.0 (Ϫ450 mV). Such a low potential redox process is unlikely to be physiologically relevant. Moreover, the reduced cluster is unstable and is rapidly degraded within minutes in the presence of excess dithionite even under strictly anoxic conditions. Similar redox behavior has been reported for the subunit bridging [4Fe-4S] cluster on IscU (36). Hence, only the oxidized form of the subunit-bridging [4Fe-4S] cluster on NfuA is likely to be relevant in vivo.
Activation of Apo-aconitase using [4Fe-4S] Cluster-loaded NfuA- Fig. 6A shows the time-dependent activation of apoaconitase that occurs when it is mixed with [4Fe-4S] clusterloaded NfuA. Initial experiments duplicated the conditions established to achieve optimal aconitase activation using [4Fe-4S] cluster-loaded IscU (29) and involved incubation of clusterloaded NfuA (12 M in [4Fe-4S] clusters) with 4 M apo-aconitase, i.e. a 3-fold stoichiometric excess of [4Fe-4S] clusters. The activation is remarkably rapid, having a second order rate constant of 6.0 Ϯ 1.5 ϫ 10 4 M Ϫ1 min Ϫ1 , a rate which is almost 10 times faster than that observed for analogous conditions using the [4Fe-4S] cluster-loaded form of IscU (29). Such rates are consistent only with transfer of intact cluster, as activation with equivalent amounts of Fe 2ϩ and S 2Ϫ ions is at least 15 times  slower. Moreover, unlike IscU-directed apo-AcnA activation, which is optimal with a 3:1 stoichiometry, NfuA-directed activation of apo-AcnA occurs with a 1:1 stoichiometry. This feature is illustrated in Fig. 6B, which shows apo-aconitase activation after 20 min of incubation as a function of the molar ratio of [4Fe-4S] clusters on NfuA to apo-AcnA. The data indicate stoichiometric [4Fe-4S] cluster transfer from NfuA to apo-AcnA. The curvature is a consequence of incomplete cluster transfer after 20 min at near stoichiometric [4Fe-4S] cluster concentrations. Indeed, the observed data are well simulated by theoretical data constructed for [4Fe-4S] cluster transfer occurring with a 1:1 stoichiometry and a second order rate constant of 6.0 ϫ 10 4 M Ϫ1 min Ϫ1 after 20 min of reaction. Hence, cluster transfer and activation of apo-aconitase is much faster and much more efficient with [4Fe-4S]-NfuA than with [4Fe-4S]-IscU.
Phenotypic Analysis of Mutant Strains Defective in NfuA Function-A strain (designated DJ1707) was constructed that has the nfuA gene partially deleted and replaced by a kanamycin gene cartridge insertion. DJ1707 has no obvious phenotypic traits with respect to growth rate when cultured in liquid medium or colony size when cultured on agar plates. However, DJ1707 cell extracts have only ϳ50% aconitase activity when compared with the isogenic wild-type strain. The diminished aconitase activity of the DJ1707 strain relative to the otherwise isogenic wild-type strain contrasted with the equivalent levels of isocitrate dehydrogenase in the two strains. Isocitrate dehydrogenase, like aconitase, is a tricarboxylic acid cycle enzyme, but it does not require a [Fe-S] cluster for its activity. Another phenotype associated with loss of NfuA function is a complete loss of capacity for growth under a 40% oxygen atmosphere (Fig. 7).
The functional importance of the two cysteine residues located within the Cys-X-X-Cys motif was tested by construct-ing strains having these residues individually substituted by alanine. These strains, DJ1759 and DJ1769, exhibit the same oxygen-sensitive growth phenotype associated with deletion and insertional inactivation of nfuA (Fig. 7).
NifU Replacement of NfuA Requires Functional NifU N-and C-terminal Domains-DJ1707 exhibited the same sensitivity to elevated oxygen concentrations whether or not NifU is also expressed. Thus, NifU is not normally able to supplant the function of NfuA. However, when nifUS expression is placed under control of the strong ara regulatory elements (Fig. 8) the oxygen-sensitive phenotype associated with loss of NfuA function   is fully corrected. Thus, there is a capacity for functional replacement of NfuA by NifU, but only when NifU is expressed at high levels. In separate experiments it was determined whether both the N-and C-terminal domains of NifU are required to rescue the phenotype associated with loss of NfuA function. This possibility was examined by construction of two different strains. One of these (DJ1773) has the NifU-Cys 35 residue substituted by alanine (DJ1773) and the other (DJ1791) has the NifU-Cys 275 residue substituted by alanine. Previous work has shown that such substitutions respectively eliminate the capacity for [Fe-S] cluster formation within the IscU-like or NfuA-like domains within NifU (8). Both strains DJ1773 and DJ1791 are also inactivated for nfuA. Results shown in Fig. 8 reveal that a capacity for the assembly of [Fe-S] clusters on both the IscU-like and NfuA-like domains of NifU is required for correction of the oxygen-sensitive phenotype associated with the loss of NfuA function.   (14 -17), IscA (11,12,37,38), and glutaredoxins (39 -41), could collectively serve as a physiological [Fe-S] cluster reservoir. This possibility is in line with the manifestation of a clear phenotype for NfuA and IscA (3) in A. vinelandii under conditions of oxygen stress, a situation that could increase the demand for [Fe-S] clusters. Also, the fact that both the N-terminal and C-terminal domains within NifU are required to replace the function of NfuA, indicates the two domains have functional interdependence. Specifically, the observed phenotype is consistent with the possibility that [Fe-S] clusters assembled on the N-terminal (IscU-like) domain within NifU can be transferred to the C-terminal (NfuA-like) domain. The converse is unlikely to occur because maturation of nitrogenase does not require an intact NfuA-like domain within NifU.

DISCUSSION
Another interesting feature of NfuA is that it shares sequence identity with the IscA family of proteins, but this sequence conservation does not include cysteine residues that are conserved among the IscA family. This feature suggests that amino acid sequence conservation between the IscA family and NfuA could be related to target specificity, rather than [Fe-S] cluster assembly, and supports the possibility that IscA and NfuA could have some overlapping functions. Given the large number of [Fe-S] proteins within the cell, as well as the variety of potential environmental conditions a cell might encounter, a hierarchy of intermediate [Fe-S] cluster carriers that could control the distribution of [Fe-S] clusters would be a reasonable strategy for maximizing metabolic capacity. There is already precedent for such a strategy in the case of E. coli, because there are at least two different systems that can function for [Fe-S] protein maturation under different conditions. One of these, the ISC system, operates under standard laboratory conditions whereas the SUF system operates under conditions of iron limitation or oxygen stress (42). It should be noted that A. vinelandii does not encode a SUF system for [Fe-S] cluster assembly. Other cases of apparent specialized targeting of preassembled [Fe-S] clusters involve the role of ErpA, an IscA-like protein, necessary for the maturation of IspG (43), and the roles of NifU and NifS in maturation of nitrogenase (44). Future studies will be aimed at testing the hypothesis that NfuA and the NfuA-like domain of NifU function as intermediate carriers of [Fe-S] clusters preformed on the respective IscU and IscU-like scaffolds.
A final important feature to emerge from this and other studies is the apparent organizational and physiological plasticity of [Fe-S] protein maturation systems from different organisms. This aspect is highlighted in Fig. 1, which compares a variety of proteins having Nfu-like modules. It is particularly noteworthy that the cluster types that can be assembled on Nfu-like modules, as well as phenotypes associated with loss of Nfu-like function, are reported to be different for different organisms. For example, in vitro reconstitution of a chloroplastic Nfu-like protein from plants has been shown to result in formation of a [2Fe-2S] cluster, and this protein is essential for maturation of chloroplastic [Fe-S] proteins (16,17). Thus, it appears that while Nfu-type modules have auxiliary or specialized functions in certain organisms, such as A. vinelandii, they have primary functions in [Fe-S] protein maturation in other organisms. Such variations in the [Fe-S] protein maturation process in different organisms are probably linked to physiological conditions, most significantly, intracellular redox conditions. Results similar to those reported here using A. vinelandii have also been found for the NfuA protein from E. coli (45).