A Bridging [4Fe-4S] Cluster and Nucleotide Binding Are Essential for Function of the Cfd1-Nbp35 Complex as a Scaffold in Iron-Sulfur Protein Maturation*

Background: The P-loop NTPases Cfd1 and Nbp35 form a scaffold complex for cytosolic-nuclear Fe-S cluster synthesis. Results: Two C-terminal cysteine residues of each Cfd1 and Nbp35 assemble a transient [4Fe-4S] cluster in a nucleotide-dependent fashion. Conclusion: The results suggest a bridging nature of the transient [4Fe-4S] cluster of Cfd1-Nbp35. Significance: This study defines the molecular role of the Cfd1-Nbp35 tetrameric scaffold complex in Fe-S cluster assembly. The essential P-loop NTPases Cfd1 and Nbp35 of the cytosolic iron-sulfur (Fe-S) protein assembly machinery perform a scaffold function for Fe-S cluster synthesis. Both proteins contain a nucleotide binding motif of unknown function and a C-terminal motif with four conserved cysteine residues. The latter motif defines the Mrp/Nbp35 subclass of P-loop NTPases and is suspected to be involved in transient Fe-S cluster binding. To elucidate the function of these two motifs, we first created cysteine mutant proteins of Cfd1 and Nbp35 and investigated the consequences of these mutations by genetic, cell biological, biochemical, and spectroscopic approaches. The two central cysteine residues (CPXC) of the C-terminal motif were found to be crucial for cell viability, protein function, coordination of a labile [4Fe-4S] cluster, and Cfd1-Nbp35 hetero-tetramer formation. Surprisingly, the two proximal cysteine residues were dispensable for all these functions, despite their strict evolutionary conservation. Several lines of evidence suggest that the C-terminal CPXC motifs of Cfd1-Nbp35 coordinate a bridging [4Fe-4S] cluster. Upon mutation of the nucleotide binding motifs Fe-S clusters could no longer be assembled on these proteins unless wild-type copies of Cfd1 and Nbp35 were present in trans. This result indicated that Fe-S cluster loading on these scaffold proteins is a nucleotide-dependent step. We propose that the bridging coordination of the C-terminal Fe-S cluster may be ideal for its facile assembly, labile binding, and efficient transfer to target Fe-S apoproteins, a step facilitated by the cytosolic iron-sulfur (Fe-S) protein assembly proteins Nar1 and Cia1 in vivo.

Fe-S clusters are ancient metal cofactors of proteins playing an essential role in many diverse biological processes. These entities are used as active sites and redox centers in enzymes, and they participate in metabolism, respiration, gene regulation, tRNA modification, DNA replication, and repair (1,2). Although Fe-S proteins can be assembled in vitro from inorganic components and apoproteins, it is well established that complex machineries are required for their in vivo assembly in both bacteria and eukaryotes (2)(3)(4)(5)(6)(7)(8)(9). One of the reasons for an orchestrated assembly inside the cell is the reactivity and toxicity of high concentrations of iron and sulfide ions. Bacteria and mitochondria of eukaryotic cells contain an evolutionary related Fe-S cluster (ISC) 2 assembly system encompassing a cysteine desulfurase (bacterial IscS and eukaryotic Nfs1-Isd11). The desulfurase delivers the sulfur removed from cysteine to Fe-S scaffold proteins (IscU and Isu1) on which iron and sulfide are assembled to a transiently bound Fe-S cluster. Further central factors of the ISC assembly machineries encompass a ferredoxin (Fdx and Yah1) for electron transfer and chaperones (HscA/HscB and Ssq1/Jac1) for Fe-S cluster transfer from the scaffold to target apoproteins (10 -12).
Studies in baker's yeast (Saccharomyces cerevisiae) and human cells have identified a role of the mitochondrial ISC assembly machinery also in the biogenesis of extramitochondrial Fe-S proteins (5,13). The connection between mitochondria and cytosolic-nuclear Fe-S protein assembly is formed by the mitochondrial inner membrane ABC transporter Atm1 that exports a compound of unknown nature to the cytosol (14,15). This compound is then used by the cytosolic Fe-S protein assembly (CIA) machinery that consists of the two related P-loop NTPases Cfd1 (16) and Nbp35 (17), Nar1, an Fe-S protein with homology to iron-only hydrogenases (18,19), the seven-bladed ␤-propeller protein Cia1 (20,21), the diflavin reduc-tase Tah18, and the Fe-S protein Dre2 (22,23). The importance of the CIA machinery is documented by the fact that all six known CIA proteins are encoded by essential genes and are required for the assembly of virtually all cytosolic and nuclear target Fe-S proteins. The only known examples of CIA-independent Fe-S protein assemblies are the cytosolic monothiol glutaredoxins Grx3/Grx4, which bind a bridging and glutathione-coordinated [2Fe-2S] cluster (24) and the CIA factor Dre2 (23). The CIA machinery appears to be functionally conserved in eukaryotes. Synthesis of the human sequence relatives of yeast Cia1, Dre2, and Tah18 complemented the depletion of the respective yeast counterparts (21)(22)(23). Furthermore, RNAimediated depletion of the human relatives of yeast Nar1 and Nbp35 in cell culture resulted in specific defects of cytosolic but not mitochondrial Fe-S proteins (25,26).
Cell biological and biochemical experiments have provided initial insights into the major steps of cytosolic Fe-S protein assembly and the potential function of Cfd1 and Nbp35. In vivo experiments with yeast cells suggested that Cfd1 and Nbp35 can assemble Fe-S clusters de novo requiring the function of, e.g. mitochondrial Nfs1-Isd11 as a sulfur donor (27) and Tah18-Dre2 for electron transfer (23). Because Nar1 and Cia1 are not required in these early steps despite their necessity for Fe-S cluster assembly on target proteins, their function was suggested to be needed later, i.e. in Fe-S cluster transfer from Cfd1-Nbp35 to target proteins. In vitro studies in fact confirmed both the binding of labile Fe-S clusters to a hetero-tetrameric complex of Cfd1 and Nbp35 and their efficient transfer to target Fe-S proteins (27). The ability to assemble and deliver Fe-S clusters both in vivo and in vitro support a scaffold role for Cfd1 and Nbp35, suggesting that their function may be similar to the mitochondrial Isu1 and bacterial IscU scaffold proteins of the ISC assembly systems (3,8). Nevertheless, our knowledge of how the labile Fe-S clusters are assembled and bound on Cfd1-Nbp35 and how they are transferred is still scarce. Moreover, the function of the nucleotide binding motif is unclear to date.
The C termini of both Cfd1 and Nbp35 contain four conserved Cys residues that were proposed to provide the scaffold function for Fe-S cluster synthesis (16,17,27,28) (Fig. 1A). Three of these four C-terminal cysteine residues (CX 18 CPXC) are conserved between Cfd1 and Nbp35, whereas the fourth occurs at non-equivalent positions. Additionally, Nbp35 possesses an N-terminal extension ( Fig. 1A) with four conserved cysteine residues (CX 13 CX 2 CX 5 C motif) that was shown to bind a [4Fe-4S] cluster (17,26,27). Assembly of this Fe-S cluster was suggested to require the function of the electron transfer chain NADPH-Tah18-Dre2 (23).
The central goal of this study was a detailed molecular characterization of Fe-S cluster binding to Cfd1 and Nbp35 with the aim of understanding which properties guarantee the ability to both assemble and transfer Fe-S clusters. To this end we functionally analyzed the roles of the nucleotide binding motif and the conserved cysteine residues in Cfd1 and Nbp35 by introducing site-directed mutations into these motifs. We then investigated the consequences of these alterations for the viability of yeast cells and Fe-S cluster assembly on Cfd1 and Nbp35 as well as for the functionality and complex formation of Cfd1-Nbp35. Our genetic, cell biological, and biochemical studies were complemented by EPR and Mössbauer spectroscopy to gain insights into the chemical nature of the bound Fe-S clusters. The results suggest a model for the mode of Fe-S cluster coordination within the Cfd1-Nbp35 complex, explaining its specific properties as a Fe-S scaffold device.

EXPERIMENTAL PROCEDURES
Yeast Strains and Cell Growth-S. cerevisiae strain W303-1A was used as wild-type (MATa, ura3-1, ade2-1, trp1 -1, his3-11,15, leu2-3,112). Previously published galactose-regulated mutants were Gal-CFD1 (27) and Gal-NBP35 (17). GalL-3HA-CFD1 and GalL-3HA-NBP35 mutant strains (abbreviated as GalL-CFD1 and GalL-NBP35 in text and figures) were constructed by homologous recombination in which the upstream promoter regions of CFD1 or NBP35 were replaced by a PCR product containing the NatNT2 resistance marker and the GALL promoter and an N-terminal triple HA tag (29). This promoter is more tightly repressed in glucose media and typically leads to a 5-fold lower expression level in galactose in comparison to the GAL1-10 promoter. Correct insertion of the regulate-able promoter was verified by PCR analysis of chromosomal DNA. Cells were grown in rich (YP) and minimal (SC) media containing the required carbon sources at a concentration of 2% (w/v) (30).
Generation of Mutants in Nucleotide Binding Site and Cysteine Residues of Cfd1 and Nbp35-To create nucleotide binding and cysteine mutant Cfd1 and Nbp35 proteins, we designed oligonucleotides (supplemental Table S1) to mutagenize CFD1 and NBP35 genes in pBluescript. After 18 PCR amplification cycles using Phusion polymerase (New England Biolabs), the parental DNA was digested using DpnI. The product was column-purified (NucleoSpin Extract II, Macherey-Nagel, Germany) and transformed into competent DH5␣ Escherichia coli cells. For more difficult amplifications, a two-step PCR was performed. In the first step, four cycles of linear amplification were carried out with single primers. The resulting products were mixed in equal proportions, submitted to 16 PCR cycles, and treated as above. Mutant genes were subcloned by PCR amplification from pBluescript into E. coli or S. cerevisiae expression vectors. Mutations were confirmed by DNA sequencing.
Bacterial Plasmids and Protein Expression in E. coli-pET-Duet-1 plasmids (Novagen) were used for heterologous expression of N-terminal-His 6 -tagged Nbp35 and Cfd1 in BL21 (DE3) E. coli host cells. Simultaneous expression of His 6 -Cfd1 and Nbp35-Strep (in multiple cloning sites 1 (MCS1) and 2 (MCS2) of pETDuet-1, respectively) was carried out with co-transformed pOFXtac SL2 (encoding E. coli GroEL) and pISC (E. coli isc operon) plasmids to enhance protein solubility. For complex formation of cysteine mutant proteins His 6 -Cfd1 with wildtype Nbp35-Strep, the coding region of wild-type Cfd1-His 6 in MCS1 was excised and replaced by a mutated copy. For protein production, a single colony harboring the three plasmids was incubated in 50 ml of LB medium with antibiotics at 37°C for 16 h under shaking (250 rpm). The preculture was diluted (1% inoculum) in 2 liters of LB medium containing the appropriate antibiotics and allowed to grow at 37°C and 170 rpm. When the culture reached an optical density of ϳ0.5 the shaker temperature was adjusted to 30°C. After cooling, FeCl 3 (50 M final concentration) was added, and overexpression was induced by the addition of isopropyl 1-thio-␤-D-galactopyranoside (1 mM final concentration). After 3 h, cells were harvested by centrifugation, washed, shock-frozen, and stored at Ϫ80°C until use. For individual expression of Cfd1 cysteine mutant proteins, cells were grown until an optical density of ϳ0.3 in the presence of 3% (v/v) ethanol. At this point, benzyl alcohol (1% (v/v) final concentration) was added, and the temperature was lowered to 20°C. After the addition of 50 M FeCl 3 , overnight induction was started by the addition of 1 mM isopropyl 1-thio-␤-D-galactopyranoside.
Protein Purification-Wild-type or mutated Cfd1 and/or Nbp35 were purified using Ni-NTA-agarose columns (Qiagen) or Strep-Tactin spin columns (IBA, Göttingen, Germany). Strep-tagged proteins were purified according to the manufacturer's instructions using a lysis washing buffer containing 100 mM Tris-HCl, pH 8.0, 150 mM NaCl supplemented with 1 mM biotin for elution.
Ni-NTA purification was initiated by resuspension of E. coli cells in lysis buffer (50 mM NaHPO 4 , 300 mM NaCl, 10 mM imidazole, 10% (v/v) glycerol, pH 8.0) and disruption at 4°C by either 1 passage at 1.0 ϫ 10 8 N/m 2 through an EmulsiFlex-C3 homogenizer (Avestin) or by sonication (3 times 30 s on ice with 1 min intervals). After centrifugation (100,000 ϫ g, 45 min, 4°C) the resulting supernatant was loaded on Ni-NTA resin pre-equilibrated with lysis buffer. The column was washed with 10 bed volumes lysis buffer including 20 mM imidazole and eluted with lysis buffer plus 250 mM imidazole. Due to the weak binding of Nbp35-His 6 to the Ni-NTA matrix, imidazole in all buffers had to be replaced by histidine at 1, 5, and 150 mM concentrations in lysis, wash, and elution buffers, respectively. For removal of salts, gel filtration of purified proteins was carried out with a PD-10 column (GE Healthcare) equilibrated with 25 mM Tris-HCl, pH 8.0, 150 mM NaCl, 10% (v/v) glycerol. The purified proteins were separated into aliquots and stored at Ϫ80°C until use. 55 Fe Incorporation into Cfd1 and Nbp35 in Vivo-For 55 Fe labeling experiments the genes encoding the Cfd1 or Nbp35 nucleotide binding and cysteine mutants were subcloned into the yeast vector pRS416-MET25 or pRS424-TDH3 (31) containing an N-terminal 3HA or a C-terminal TAP-tag sequence, respectively (32). In vivo radiolabeling of transformed yeast cells with 55 FeCl 3 and measurement of 55 Fe incorporation into 3HA or TAP-tagged Cfd1 and Nbp35 via anti-HA agarose or IgG-Sepharose immunoprecipitation was carried out as described (33).
Growth Complementation and Leu1 Activity Measurement in S. cerevisiae-Growth complementation was tested using GalL-CFD1 or GalL-NBP35 yeast cells transformed with pRS416 harboring wild-type or cysteine-mutated versions of CFD1 and NBP35 under the control of their endogenous promoters. For this purpose, the SacI-XbaI MET25 promoter fragment in pRS416-MET25 (31) was replaced by a 500-bp PCR fragment corresponding to the CFD1 or NBP35 promoter region. Subsequently, the coding region of CFD1, NBP35, or the respective mutated versions in pBluescript were amplified with primers that allowed cloning into the XbaI-BamHI (CFD1) or SpeI-BamHI (NBP35) restriction sites. After transformation of the plasmids into GalL-CFD1 or GalL-NBP35 cells, growth in liquid minimal media supplemented with 2% galactose or glucose was for 40 h. Aliquots of a 5-l cell suspension diluted to an optical density 0.5 at 600 nm and consecutive 10-fold serial dilutions were spotted on agar plates containing minimal media supplemented with 2% galactose or glucose. Plates were incubated for 2 days at 30°C and photographed. Results were repeated at least three times by independent transformations. Leu1 activity was determined in extracts prepared from cells grown as described above (27).
Protein-Protein Interaction Studies-To examine the ability of yeast Cfd1 and Nbp35 or respective cysteine mutant proteins to interact with itself or to form a complex with the partner protein, the corresponding genes were cloned into yeast expression vectors encoding HA or Myc-epitope tags (supplemental Table S2) at their N termini. The correctness of the resulting HA-Cfd1, Myc-Cfd1, HA-Nbp35, and Myc-Nbp35 sequences was confirmed by DNA sequencing. These plasmids were transformed into Gal-CFD1 or Gal-NBP35 yeast cells according to the association to be studied. The resulting transformants were depleted for endogenous Cfd1 or Nbp35 by growth on minimal medium supplemented with glucose for 40 h. Input and Myc-or HA-immunoprecipitated proteins were prepared as described (23) and analyzed by Western blot after separation by SDS-PAGE.
Chemical Reconstitution of Fe-S Clusters on Cfd1 and Nbp35-Cfd1 (wild-type and cysteine mutants), Nbp35, and the Cfd1-Nbp35 complex were reduced at a protein concentration of 20 M in an anaerobic chamber (Coy Laboratory Products, Ann Arbor, MI) with 5 mM DTT for 1 h at 25°C in reconstitution buffer (25 mM Tris-HCl, pH 8.0, 150 mM NaCl). Anaerobic stock solutions of ferric ammonium citrate and Li 2 S (10 mM in water) were prepared freshly. Reconstitution of Cfd1 was started by the addition of 100 M ferric ammonium citrate with stirring. After 5 min 100 M Li 2 S was slowly added. In a similar fashion 2-or 3-fold higher iron and sulfur concentrations were used for Nbp35 or the Cfd1-Nbp35 complex, respectively. Reconstituted proteins were desalted after 3 h on a PD-10 column equilibrated with reconstitution buffer. Incorporation of the Fe-S clusters into apoproteins was monitored by UV-visible spectroscopy (V-550, Jasco, Inc.).
EPR and Mössbauer Spectroscopy-Fe-S proteins were chemically reconstituted for EPR spectroscopy as described above. The individual Cfd1 and Nbp35 proteins could not be reconstituted at protein concentrations higher than 1-2 mg/ml. Therefore, we reconstituted them at low protein concentrations followed by desalting and Millipore Ultra-4 Ultracel 30-kDa spin column membrane concentration under anaerobic conditions. However, this method did not yield high enough concentrations for high spin EPR analysis. For Cfd1-Nbp35 complex highly concentrated samples (ϳ0.7 mM protein) could be obtained. After the addition of 5 mM MgCl 2 and 2.5 mM neutralized ATP or GTP if applicable, the Fe-S clusters were reduced with sodium dithionite (4 mM final concentration), and samples were shock-frozen after 2 min. EPR spectra were recorded at cryogenic temperatures with a Bruker ESP 300E X-band spectrometer equipped with an Oxford Instruments ESR910 helium flow cryostat. The microwave frequency was determined with a Hewlett-Packard 5340A frequency counter. Spin quantification of the S ϭ 1 ⁄ 2 and Boltzmann population corrected isotropic high spin resonances was carried out under non-saturating conditions of temperature and microwave power using a Cu 2ϩ standard solution in 2 M NaClO 4 and 10 mM HCl. The g values and energy separations of the doublets of the high spin systems were calculated with the program Rhombo (34) or for S ϭ 11/2 with a program written by E. Bill (available upon request).
For Mössbauer studies Cfd1, Nbp35, and the complex were reconstituted anaerobically with the 57 Fe isotope. 57 FeSO 4 was prepared by boiling metallic 57 Fe in 1 M H 2 SO 4 in a silicon oil bath. After cooling 1 eq of sodium citrate, per 57 Fe was added, and the solution was adjusted to ϳpH 6 by the slow addition of anaerobic 2.7 M ammonia solution to yield 57 Fe-labeled ferric ammonium citrate. The iron concentration was colorimetrically determined with ferene (27). The reconstituted samples (after desalting) were concentrated to about 0.2 mM to avoid precipitation. Mössbauer samples were frozen in the anaerobic chamber by placement on an aluminum block precooled in liquid nitrogen outside the anaerobic chamber and kept at Ϫ80°C until analysis. Mössbauer spectra were recorded and simulated as described (35).
Protein, Iron, and Sulfide Quantification-Protein content was determined by quantitative amino acid analysis after hydrolysis in 6 M HCl by PANATecs GmbH (Tübingen, Germany). Concentrations of the most stable amino acids (Ala, Gly, AspϩAsn, GluϩGln, Leu) were divided by the number of residues per protein or protein complex (including tags). With a relative standard deviation of 3%, all seven amino acids used yielded an identical protein concentration. The analysis also showed that the stoichiometry of the His 6 -Cfd1-Nbp35-Strep complex was 0.95 Ϯ 0.09. Protein determinations were carried out with the micro-Biuret method after trichloroacetic acid/ desoxycholate precipitation or the Bio-Rad protein assay using BSA as a standard, and the same preparations were analyzed by quantitative amino acid analysis. Non-heme iron and sulfide contents of reconstituted samples were determined as described (18).

Identification of Cysteine Residues of Cfd1 and Nbp35 with
Importance for Cell Growth-To study the functional role of residues potentially responsible for Fe-S cluster coordination in Cfd1 and Nbp35, conserved cysteine residues in these proteins were changed to alanine using site-directed mutagenesis of the encoding genes. In addition to single exchanges a number of double mutations were generated (Fig. 1). We first tested the functional importance of the various cysteine residues of Cfd1 and Nbp35 by complementing the regulate-able GalL-CFD1 and GalL-NBP35 yeast strains with the mutated CIA genes. These strains contain a truncated GAL1-10 promoter that leads to production of moderate levels of protein in the presence of galactose (29). In the presence of glucose, gene expression is strongly decreased. The plasmid-encoded CIA component (i.e. mutant Cfd1 and Nbp35) was maintained at close to physiological protein levels by expression from centromeric plasmids under the control of their endogenous promoters. GalL-CFD1 and GalL-NBP35 cells containing these plasmids were grown on galactose-or glucose-containing minimal medium, and cell growth was tested. The Cfd1 cysteine single mutant proteins C182A and C207A or the double mutant C182A/C207A were able to fully complement the growth retardation of depleted GalL-CFD1 cells, whereas single or double mutations in the two central cysteine residues (Cys-201 and Cys-204) did not improve growth (Fig. 1A and Ref. 16). The situation for the C-terminal cysteine residues of Nbp35 was remarkably similar to that of Cfd1. For growth of the GalL-NBP35 cells on glucosecontaining medium, the central cysteine pair of Nbp35 (Cys-253 and Cys-256) was essential, whereas the proximal cysteine residues at positions 234 and 295 were dispensable (Fig. 1B). We conclude that only two of four C-terminal cysteine residues of Cfd1 and Nbp35 are required for growth, identifying them as potential ligands for the C-terminal Fe-S cluster. We next tested the functional importance of the four N-terminal cysteine residues of Nbp35 (Fig. 1B). Three of the four N-terminal cysteine residues of Nbp35 were essential for growth of Nbp35-depleted cells, whereas mutation of the first cysteine residue (Cys-27) to alanine supported growth, yet at a slower rate. Together, these results show that three N-terminal cysteine residues of the [4Fe-4S] binding motif of Nbp35 are likely Fe-S cluster-coordinating ligands, whereas the role of the first cysteine residue remains elusive.
C-terminal Central Cysteine Residues of Cfd1 and Nbp35 Are Fe-S Cluster Ligands in Vivo-To investigate the importance of the cysteine residues in Cfd1 and Nbp35 for Fe-S cluster coordination, yeast cells were radiolabeled with 55 Fe, and Cfd1 and Nbp35 were immunoprecipitated (27,33). Due to the low abundance of these proteins (estimated to be Ͻ50 molecules per cell (36)) 55 Fe incorporation was performed using plasmids harboring TAP-or HA-tagged versions under the control of the TDH3 or MET25 promoter. In vivo radiolabeling with 55 Fe showed that the extent of Fe-S cluster binding to C182A, C207A, and C182A/C207A mutant Cfd1 was similar to wild-type controls (Figs. 2A and supplemental Fig. S1). Thus, the proximal residues Cys-182 and Cys-207 play no critical role in Fe-S cluster binding to Cfd1. In contrast, 55 Fe-S cluster binding to C201A and C204A mutant HA-Cfd1 was severely affected, whereas the protein levels remained similar to wild type ( Fig. 2A). Notably, the inability to stably bind an Fe-S cluster resulted in severe degradation of the C201A, C204A, and C201A/C204A Cfd1-TAP-mutant apoproteins, as much weaker signals were detected by immunoblotting supplemental Fig. S1). The low stability of apoproteins is a well known phenomenon (18,19). These results were fully corroborated by measuring the functionality of Cfd1 (mutant) proteins in the maturation of the cytosolic [4Fe-4S] cluster-containing enzyme isopropyl malate isomerase, Leu1 (14,33). This method does not require overexpression of CFD1 and NBP35. If the CIA machinery is functional, the Leu1 enzyme activity typically is ϳ200 milliunits/mg in yeast cell extracts in wild-type, GalL-CFD1, or GalL-NBP35 cells ( Fig. 2B and Refs. 16 and 17). Depletion of Cfd1 hampers the incorporation of the Fe-S cluster into Leu1 and thus greatly impairs its activity (Յ30 milliunits/mg). Nearly wild-type Leu1 activities were detected for the C182A, C207A, and C182A/ C207A mutant Cfd1 synthesized in Cfd1-depleted cells. In contrast, low activities, close to background levels, were measured for C201A, C204A, and C201A/C204A mutant Cfd1. In all experiments the Leu1 protein amounts determined by immunoblotting remained constant (Fig. 2B, lower panel). Taken together, growth complementation, Cfd1 55 Fe incorporation, and Leu1 activity measurements strongly suggested the participation of the central residues Cys-201 and Cys-204 but not of Cys-182 and Cys-207 as Fe-S cluster ligands in Cfd1.
Similar measurements were performed for mutations of the conserved N-and C-terminal cysteine residues of Nbp35. For mutations of the N-terminal Fe-S cluster binding motif of Nbp35, both 55 Fe incorporation and Leu1 enzyme activity data unequivocally showed the involvement of the Cys-41, Cys-44, and Cys-50 residues in both 55 Fe-S cluster binding and Leu1 enzymatic function (Fig. 3). Consistent with the growth complementation experiments (Fig. 1B), mutation of residue Cys-27 yielded intermediate effects in both assays, showing that this residue is of importance for both Fe-S cluster coordination and function of Nbp35, yet not absolutely required (Fig. 3A). Because some C-terminal TAP-tagged Nbp35 mutant proteins were prone to degradation in these experiments (not shown), we used HA-tagged versions that were stable. Upon exchange of the central cysteine residues (C253A, C256A, and C253A/ C256A), 55 Fe-S cluster binding to Nbp35-HA was strongly affected (Fig. 3B), and Leu1 activities were diminished to background levels in Nbp35-depleted cells (Fig. 3C). In fact, the complete loss of 55 Fe binding by Nbp35 with these central cysteine mutations and of the N-terminal mutations demonstrated that the two Fe-S clusters mutually affect their stability, sug- A, an empty 416 plasmid or plasmids encoding wild-type (WT) or the indicated mutated HAtagged Cfd1 proteins under the control of the MET25 promoter were transformed into yeast strain Gal-CFD1. Cells were cultivated in galactose-containing minimal medium for 24 h. Incubation was continued for 16 h in iron-poor minimal medium followed by radiolabeling with 55 Fe for 2 h. After preparation of cell extracts with glass beads and affinity isolation of HA-Cfd1, the amounts of Cfd1-bound 55 Fe were measured by scintillation counting. B, an empty 416 plasmid or the same plasmid containing untagged WT or cysteine mutant versions of CFD1 under control of the endogenous promoter were transformed into W303 (wild type) or GalL-CFD1 strains. Cells were grown in glucose-containing minimal medium for 40 h, cell extracts were prepared using glass beads, and the isopropyl malate isomerase activity of Leu1 was immediately measured. The lower panels in A and B show immunostains of cell extracts visualizing the amounts of HA-Cfd1 and Leu1, respectively.
gesting that their assembly and/or stability is cooperative. Because the Leu1 activities and Nbp35 55 Fe incorporation for C234A and C295A variants were comparable with those of wild-type Nbp35, these results suggested that only the C-terminal central cysteine residues Cys-253 and Cys-256 of Nbp35 were critical for functionality and Fe-S cluster binding to the C terminus.
Central C-terminal Cysteine Residues Are Required for Cfd1-Nbp35 Complex formation-Upon recombinant expression and purification, Cfd1 and Nbp35 appear as a stable, heterotetrameric complex (27). The interaction between Cfd1 and Nbp35 is also observed in vivo. We tested the importance of the conserved cysteine residues of Cfd1 and Nbp35 for complex formation. To this end His 6 -Cfd1 (wild-type or cysteine mutants) was co-expressed together with wild-type Nbp35-Strep in E. coli from a pETDuet-1 vector, yielding similar levels of these overproduced, tagged proteins (Fig. 4, left; compare non-induced and induced samples; asterisks). The cell extracts  were loaded onto Streptactin affinity columns, and protein specifically bound to the resin was eluted (Fig. 4, right). Although the amount of purified Nbp35-Strep was similar under all conditions, the amount of co-eluting His 6 -Cfd1 depended on the mutation of Cfd1. We confirmed that Cfd1 proteins could be purified by Ni-NTA chromatography, and thus the lack of copurification with Nbp35 reflected the absence of complex formation. The C182A and C207A exchanges in Cfd1 had no marked effect on Cfd1-Nbp35 complex formation, whereas mutations of cysteine 201 or 204 completely eliminated formation of a stable complex. Taken together, these results strongly suggested a critical role of the two central cysteine residues of Cfd1 for complex formation.
The physiological importance of the four conserved cysteine residues of Cfd1 and Nbp35 for in vivo interaction with itself or for Cfd1-Nbp35 complex formation was investigated further by anti-HA-tag immunoprecipitation from yeast cell extracts. Our results suggested that both wild-type Cfd1 and Nbp35 interacted with themselves, as both HA-Cfd1/Myc-Cfd1 and HA-Nbp35/Myc-Nbp35 pairs were co-immunoprecipitated (Fig. 5, A and B). Cfd1-Cfd1 interaction was not dependent on any of the C-terminal cysteine residues (Fig. 5A), whereas the central cysteine pair (C253A and C256A) was critical for Nbp35-Nbp35 interaction (Fig. 5B). Apparently, the Nbp35-Nbp35 interaction was dependent on the two Fe-S cluster-coordinating residues, although Cfd1 stably interacted with itself in the absence of bound Fe-S cluster. We then investigated the heterologous interaction between HA-Cfd1 (mutant) proteins and Myc-Nbp35 by anti-HA-tag immunoprecipitation in vivo. Cfd1-Nbp35 co-immunoprecipitation was efficient and depended on residues Cys-201 and Cys-204 of Cfd1 (Fig. 5C). Likewise, when co-expressed HA-Nbp35 (mutant) proteins and Myc-Cfd1 were analyzed by co-immunoprecipitation, the interaction of both proteins was dependent on residue Cys-253 of Nbp35 (Fig. 5D). We conclude from these results that Cfd1 and Nbp35 interact in vivo in a fashion that is compatible with the hetero-tetrameric complex identified in vitro (27). In most cases the complex stability depended on the two central C-terminal Cys residues of Cfd1 and Nbp35, thus suggesting a critical role of these residues and in turn Fe-S cluster binding, for complex formation. This result is compatible with the idea of a bridging Fe-S cluster that appears to stabilize the protein interactions within the Cfd1-Nbp35 complex.
Mössbauer Spectroscopy of Reconstituted Cfd1, Nbp35, and Complex Detects [4Fe-4S] 2ϩ Clusters-The type of Fe-S cluster bound to Cfd1-Nbp35 was investigated by Mössbauer spectroscopy. The cluster composition in the diamagnetic (EPR-silent) form was analyzed after reconstitution with 57 Fe and sulfide. UV-visible spectroscopy confirmed that these samples had properties similar to those described previously (27). The Mössbauer spectra of Cfd1, Nbp35, and the complex recorded at 80 K (Fig. 6, A-C) were dominated by the quadrupole doublet of a main component (I) with an isomer shift (␦) of 0.43 mm s Ϫ1 and quadrupole splitting (⌬E Q ) of 1.18 mm s Ϫ1 , values typical for all cysteine-coordinated [4Fe-4S] 2ϩ clusters (35). For the complex the diamagnetic nature of component I was supported by the spectrum recorded at 4.2 K with a magnetic field of 6.5 Tesla (Fig. 6D). An excellent fit to the experimental data was obtained with ␦ ϭ 0.45 mm s Ϫ1 , ⌬E Q ϭ 1.19 mm s Ϫ1 , and an asymmetry parameter of 0.70. In this spectrum the second component (II, ␦ ϭ 0.34 mm s Ϫ1 and ⌬E Q ϭ 0.62 mm s Ϫ1 ) with a relative contribution of 25% to the spectrum of the complex recorded at 80 K in the absence of applied field (Fig. 6C)  ) with a relative contribution of 12, 6, and 6% in Cfd1, Nbp35, and the complex, respectively, originated from non-sulfur-coordinated high spin ferrous iron, a species commonly observed in Mössbauer spectra of reconstituted Fe-S proteins (35,37). Components II and III make up 46% of the total iron in Cfd1 (34% plus 12%, respectively), 36% in Nbp35 (30% plus 6%, respectively), and 31% in the complex (25% plus 6%, respectively). The relative content of non-Fe-S components II and III determined by deconvolution was used to calculate the net iron stoichiometry per protein using the total bound iron content as determined previously (27) and was corrected for components II/III. The corrected stoichiometry of iron per protein was 2.5 Ϯ 0.4 (Cfd1), 5.2 Ϯ 1.0 (Nbp35), and 8.7 Ϯ 0.8 (Cfd1-Nbp35 heterodimer). We confirmed by quantitative amino acid analyzes that the stoichiometry was not influenced by the method of protein determination. These corrected values were compatible with a bridging [4Fe-4S] cluster between the essential CPXC motifs of Cfd1-Nbp35 and a single [4Fe-4S] cluster bound to the N terminus of Nbp35 (see "Discussion"). In this case the expected Fe-S content for Cfd1 is 2 (one [4Fe-4S] per Cfd1 dimer) and for Nbp35 is 6 (two [4Fe-4S] in the N termini plus one bridging [4Fe-4S] per Nbp35 dimer), whereas in the Cfd1-Nbp35 dimer 8 iron and 8 sulfur are expected (two [4Fe-4S] in the N termini of Nbp35 and two [4Fe-4S] bridging per Cfd1-Nbp35 hetero-tetramer). These contents per protein dimers (2, 6, and 8) match the corrected experimental values and support a bridging situation for Fe-S cluster binding at the C termini.
Evidence for High Spin EPR Signals of [4Fe-4S] 1ϩ Clusters of Cfd1-Nbp35 Complex-By double integration of the EPR signals of two axial EPR signals from S ϭ 1/2 spin states in the g ≈ 2 region, only 15-20% of the [4Fe-4S] 1ϩ clusters of dithionitereduced reconstituted recombinant Cfd1-Nbp35 complex can be detected (Fig. 7A and Ref. 27). The missing spin intensity (80 -85%) of the S ϭ 1/2 species cannot be accounted for by insufficient reduction by dithionite because Cfd1 and Nbp35 exhibit about 50% bleaching at 400 nm. A probable explanation could be that the [4Fe-4S] 1ϩ clusters are present as a physical mixture of S ϭ 1/2 and high spin states. Such physical spin mixtures are well characterized for the nitrogenase iron protein and activators of 2-hydroxyacyl-CoA dehydratases, which contain bridging [4Fe-4S] clusters (38,39). At the 50 M concentrations of Cfd1, Nbp35, or complex used previously (27), high spin states were not detected due to the low intensity and the broad nature of the EPR signals associated with such states (see Ref. 34). We, therefore, performed EPR spectroscopy using a more concentrated sample of Cfd1-Nbp35 complex (0.7 mM). Several signals at low field were observed in addition to the known intense signals at g ≈ 2 from the low spin [4Fe-4S] 1ϩ clusters (Ref. 27; Fig. 7A, cropped in Fig. 7B) and a signal at g ϭ 4.3 from adventitiously bound ferric ions. A broad absorptionshaped band at g Ϸ 6 to 4 and two sharp derivative-shaped features at g ϭ 5.15 and 7.26 were detected at 4.2 K (Fig. 7B). The intensity of the latter two signals increased upon recording the EPR spectra at 10 K (Fig. 7B, top trace). The large g values and the increase of intensity of the EPR features at higher tem-perature showed that these signals originated from EPR transitions within excited state doublets of a high spin state of the [4Fe-4S] 1ϩ clusters in Cfd1 and Nbp35. Based on this information we could determine the spin state and Hamiltonian parameters with the help of rhombograms (Ref. 34 and a program by author E. Bill (available upon request)) (Fig. 7C). Three high spin states of the [4Fe-4S] 1ϩ clusters could thus be assigned: the broad signal at g ϭ 6 to 4 is from a S ϭ 3/2 system with ͉D͉ Ͻ 1 cm Ϫ1 and E/D Ͼ 0.25, the isotropic g ϭ 5.15 signal is from the ͉Ϯ 3/2Ͼ doublet of a S ϭ 7/2 system with D ϳ 2 cm Ϫ1 and E/D ϭ 0.11, and the isotropic g ϭ 7.26 signal is from the ͉Ϯ 5/2͘ doublet of a S ϭ 11/2 system with D ϳ 1 cm Ϫ1 and E/D ϭ 0.195 (see Fig. 7B and supplemental Fig. S2 for rhombograms). The S ϭ 7/2 and S ϭ 11/2 forms represent a minority species (each about 2%, based on double integration with correction for the Boltzmann population of the EPR active doublet). We suggest that the broad S ϭ 3/2 signal at g ϭ 6 to 4, although moderate in intensity compared with the S ϭ 1/2, S ϭ 7/2, and S ϭ 11/2 EPR signals (34), accounts for the main part of [4Fe-4S] 1ϩ clusters. Identification of which EPR signal belongs to the N-terminal cluster of Nbp35 or to the bridging cluster(s) is not possible because individually expressed Cfd1 or Nbp35 could not be obtained in concentrations required for detection of EPR signals of the high spin state(s) of the clusters. In conclusion, the [4Fe-4S] clusters in Cfd1 and Nbp35 in the reduced states are present in the form of a physical spin mixture, which is commonly observed for bridging [4Fe-4S] clusters. This interpretation is corroborated by Mössbauer spectroscopy.

Importance of ATP Binding Motif of Cfd1 and Nbp35-No
role of nucleotide binding has been defined so far for the P-loop NTPases Cfd1 and Nbp35. For the nitrogenase iron protein, another P-loop ATPase (38,40), the addition of nucleotides changes the rhombicity and spin state of its bound [4Fe-4S] 1ϩ clusters. Therefore, EPR spectra of the Cfd1-Nbp35 complex were recorded in the presence of Mg 2ϩ /ATP or Mg 2ϩ /GTP. The addition of nucleotides, however, did not significantly influence the g ϭ 2 region (Fig. 7A) and the high spin EPR characteristics at 4.2 K (Fig. 7B) and higher temperatures (not shown). These findings were in full agreement with several observations we made during the biochemical investigation of recombinant Cfd1 and Nbp35. Cfd1 and Nbp35, individually or in complex, did not show any ATPase or GTPase activity (i.e. Ͻ 10 milliunits /mg) and did not bind ADP, ATP, or GTP (Ͻ 0.05 mol/mol; not shown). Despite the lack of in vitro evidence for nucleotide binding to Cfd1 and Nbp35, the Walker A nucleotide binding motifs were essential for in vivo function of both proteins. First, yeast cells bearing only the Walker A-mutated Cfd1 (K31A) or Nbp35 (K86A), generated by plasmid shuffle, were not viable (28,41). Second, maturation of the target Fe-S protein Leu1 was not supported by the mutated proteins as shown by following 55 Fe-S cluster incorporation into Leu1 (Fig. 8, A and B). As a control, plasmid-driven synthesis of wild-type Cfd1 or Nbp35 in the respective depleted strains was able to (at least partially) restore Fe-S cluster incorporation into Leu1. These findings show that the Walker A motif is functionally important but leave open what the exact role of nucleotide binding/hydrolysis in Cfd1-Nbp35 may be. A clue to this open question came from studies measuring the in vivo 55 Fe-S cluster binding to wild-type and Walker A-mutated versions of Cfd1 or Nbp35. We found that the mutated Cfd1 and Nbp35 proteins can bind Fe-S clusters almost as well as the wild-type proteins (supplemental Fig. S3). Intriguingly, Fe-S cluster incorporation into mutated Cfd1 and Nbp35 proteins was severely decreased upon depleting the wild-type copies in the respective regulated yeast strains (Fig. 8, C and D), demonstrating that Fe-S cluster assembly on mutated Cfd1 and Nbp35 depended on the presence of the respective wild-type copies in trans. Apparently, the wild-type copies of Cfd1 and Nbp35, even though present at comparatively lower levels than the plasmid-expressed mutated versions, are essential for loading Fe-S clusters onto the mutated proteins. We conclude from these results that nucleotide binding/hydrolysis seems to be critical for the load-ing of Fe-S clusters onto Cfd1 and Nbp35 in vivo. The impairment of Fe-S cluster assembly on Cfd1-Nbp35 may explain why the mutation of the Walker A motif is essential for cell viability and function of these proteins in cytosolic-nuclear Fe-S cluster assembly.

DISCUSSION
In this work we investigated the mode and type of Fe-S cluster binding to the CIA components Cfd1 and Nbp35, which form a scaffold complex for cytosolic Fe-S cluster synthesis. We defined the chemical nature of these cofactors by EPR and Mössbauer spectroscopies and demonstrated that reconstituted Cfd1 and Nbp35 bind only clusters of the [4Fe-4S] type. Further in vivo and in vitro experiments identified an essential function of two central cysteine residues in the C termini of both Cfd1 and Nbp35 for cell viability and cytosolic-nuclear Fe-S protein assembly. More specifically, the two central cysteine residues were found to be required for both Fe-S cluster coordination and stabilization of the protein interaction between Cfd1 and Nbp35. We suggest that this specific contact occurs via bridging [4Fe-4S] clusters between Cfd1 and Nbp35 homo-and/or heterodimers (Fig. 9). This assumption is corroborated by the stoichiometry of iron and sulfur binding and by the occurrence of EPR-detectable high spin states of the reduced cluster, a feature that the Cfd1-Nbp35 complex shares with several proteins containing bridging [4Fe-4S] clusters (38 -40, 42). Moreover, our studies provide evidence for the role of nucleotides in these two P-loop NTPases. In vivo experiments with Cfd1 or Nbp35 mutated in the nucleotide binding domain showed that the incorporation of Fe-S clusters into mutated Cfd1 or Nbp35 is only possible if wild-type copies of Cfd1 and Nbp35 are present in trans, suggesting that nucleotide binding and possibly hydrolysis may facilitate Fe-S cluster loading onto these scaffold proteins. Binding of nucleotides, however, seems to be too transient to be detectable by standard biochemical methods. Together, these findings identify the cysteine residues used for generating bridging [4Fe-4S] clusters on Cfd1-Nbp35 and suggest that energy input is needed for the scaffold function of Cfd1-Nbp35 in Fe-S cluster biogenesis.
Previously, Cfd1 was shown to harbor a single labile Fe-S cluster, whereas Nbp35 binds both a stable and a labile cluster (27). These assumptions were derived from the observation that Cfd1 purified from E. coli was almost colorless, although both Cfd1 and Nbp35 specifically bound a Fe-S cluster in vivo. Furthermore, the N-terminal-truncated Nbp35 (lacking the N-terminal [4Fe-4S] cluster) was also isolated in the apo form, and similar to Cfd1, a labile [4Fe-4S] cluster could be chemically reconstituted onto the apoprotein. Bioinformatic analyses showed the presence of four highly conserved C-terminal cysteine residues in both proteins, suggesting a role of these residues in Fe-S cluster coordination (27). However, no dedicated biochemical investigation on the function of these residues has been presented so far. Here, we used site-directed cysteine mutagenesis to dissect the functional importance of each of the conserved cysteine residues of Cfd1 and Nbp35 by in vivo and in vitro experiments. First, all four cysteine residues at the N-terminal extension of Nbp35 appear to be required for 55 Fe incorporation and Leu1 activity, suggesting that all these residues FIGURE 7. No effects of ATP and GTP on the EPR signals from the dithionite-reduced Fe-S clusters of the Cfd1-Nbp35 complex. A, EPR spectra were recorded at a microwave power of 2 milliwatts at 10 K. The buffer contained 4 mM MgCl 2 (all cases) and ATP or GTP (2 mM) as indicated. Samples were frozen 2 min after the addition of 4 mM sodium dithionite. B, procedures were as in A but with a microwave power of 20 milliwatt at 4 K (except for the top trace, 10 K). The intense signals at g ϭ 2 have been clipped off. Other EPR parameters were: amplitude modulation, 1 millitesla; microwave frequency, 9.46 GHz; modulation frequency, 100 kHz. C, shown is a rhombogram of the ͉Ϯ 5/2͘ multiplet for a S ϭ 11/2 system. The rhombicity, which gives rise to the sharp isotropic g ϭ 7.26 EPR signal, is highlighted by an arrow.
serve as Fe-S cluster ligands. Although residues Cys-41, Cys-44, and Cys-50 are essential for both Fe-S cluster binding and function of Nbp35, mutation of residue Cys-27 elicits a compara-tively weak phenotype. We, therefore, cannot fully exclude that the effects of the Cys-27 mutation are indirect and e.g. are caused by a conformational change that affects the stability of the N-terminal cluster. However, with the exception of His-26 whose mutation to Ala had no effect on growth (not shown), there are no other appropriate, conserved residues in the N-terminal domain that could serve as a potential fourth ligand. Hence, it seems likely that all four cysteine residues coordinate the N-terminal [4Fe-4S] cluster of Nbp35.
Bioinformatic analysis of Cfd1 and Nbp35 reveals the conservation of four cysteine residues at the C terminus. Although the first three residues are conserved in both proteins, the fourth residue is located in non-equivalent positions. Single or double mutations of the first and fourth cysteine residues of Cfd1 (Cys-182, Cys-207) and Nbp35 (Cys-234, Cys-295) to alanine did not affect Fe-S cluster binding of these proteins. In contrast, mutation of the two central cysteine residues of Cfd1 (Cys-201, Cys-204) and Nbp35 (Cys-253, Cys-256) abolished Fe-S cluster binding. Moreover, only these central cysteine residues were crucial for cell viability and function of Cfd1 and Nbp35 in the  55 Fe incorporation into endogenous Leu1 was followed as described in Fig. 2A for Gal-CFD1 cells (A) and Gal-NBP35 (B) cells grown in galactose-or glucose-containing minimal medium as indicated. C, 55 Fe incorporation into plasmid-encoded Cfd1-TAP or its K31A mutant version was measured in Gal-CFD1 cells grown in galactose-or glucose-containing minimal medium as indicated. D, 55 Fe incorporation into plasmid-encoded Nbp35-TAP or its K86A mutant version was estimated in Gal-NBP35 cells as described in C. In C and D the background values (around 2 ϫ 10 3 cpm/g cells) obtained with empty plasmids were subtracted. Immunostaining of Leu1, Cfd1, and Nbp35 in cell extracts is shown at the bottom of the panels. For all panels 424 plasmids under the control of the TDH3 promoter were used. assembly of cytosolic Fe-S proteins such as Leu1, assigning an essential physiological function in Fe-S protein biogenesis to these two residues. Sequence analysis shows that the central CPXC motif is present in virtually all eukaryotic Cfd1 and Nbp35 proteins (see Ref. 43). Because eukaryotic Cfd1 and Nbp35 are highly conserved proteins (Ͼ50% amino acid identity), it is not astonishing that the distal cysteine residues are present in most eukaryotes, including humans. However, in some fungi, such as Aspergillus (7 species), Pichia pastoris, Cryptococcus neoformans, and Magnaporthe grisea, Cfd1 and Nbp35 lack the first or fourth cysteine residues. These natural Cfd1 and Nbp35 variants provide independent evidence for the nonessential character of the distal cysteine residues in the eukaryotic CIA machinery despite their high conservation. The importance of the central cysteine pair is similar to that described for ApbC, a maturation factor for tricarballylate reductase in eubacteria (44) and for the mitochondrial complex I assembly protein Ind1 (45). However, comparison of these P-loop NTPase homologues with Cfd1 and Nbp35 has a limited functional relevance for the eukaryotic CIA machinery because ApbC and Ind1 are believed to function as homodimers, and are specific rather than general components for the maturation of target Fe-S proteins.
Our proposal of a bridging [4Fe-4S] cluster, which is bound by the two central C-terminal cysteine residues, is based on several lines of evidence including 55 Fe incorporation, functional importance for Leu1 activity, Cfd1-Nbp35 complex formations observed by affinity purifications, and isolation of Cfd1-Nbp35 complexes after expression in E. coli. All these studies demonstrated that the bridging Fe-S cluster is crucial yet not essential for stabilization of the Cfd1-Nbp35 heterotetrameric complex. Mutation of the residues binding the bridging Fe-S cluster disrupted complex formation, with one exception (Nbp35-C256A), where binding still occurred. This shows that these residues and the bound Fe-S cluster are of importance but are not the only determinant for complex formation. This notion is supported by our spectroscopic studies (discussed below) strongly suggesting the presence of a bridging [4Fe-4S] cluster that is coordinated by the central C-terminal residues of Cfd1 and Nbp35 in the heterodimer. Moreover, the stoichiometry of iron binding to the isolated complex (corrected for adventitiously bound Fe) supports the bridging nature of a [4Fe-4S] cluster. Taken together, several lines of evidence suggest that Cfd1 and Nbp35 bind a bridging [4Fe-4S] cluster at their C termini and that the lability of binding this cluster may be ideal for its transfer from the scaffold complex toward apoproteins.
Examination of purified, reconstituted Cfd1, Nbp35, and the Cfd1-Nbp35 complex by Mössbauer spectroscopy showed that the bound Fe-S clusters are of the [4Fe-4S] 2ϩ type. No signs for a [2Fe-2S] cluster were obtained. In the reduced state the clusters in the Cfd1-Nbp35 complex exhibit three different high spin EPR signals (S ϭ 3/2, 7/2 and 11/2 states) in addition to the known S ϭ 1/2 signals (27). Although high spin states were once thought to be uncommon for [4Fe-4S] 1ϩ clusters, several proteins harboring a mixture of S ϭ 1/2 and S Ն 3/2 high spin states or with several high spin states are known. Notably, in the majority of these proteins the [4Fe-4S] cluster bridges the two subunits. In some proteins such as benzoyl-CoA reductase (46), iron-only hydrogenase maturation protein HydF (47), and the transcriptional regulator SufR (48), the bridging nature has been postulated or seems likely. For crystallographically characterized proteins (49 -51) high spin states of a bridging [4Fe-4S] cluster occur in the nitrogenase iron protein (38,40), the activator of 2-hydroxyglutaryl-CoA dehydratase (39), and the F X cluster of the photosynthetic reaction center (42). Cfd1-Nbp35 now join the increasing inventory of [4Fe-4S] proteins with a bridging Fe-S cluster and a high spin state in the reduced form.
Cfd1 and Nbp35 belong to the P loop NTPases of the SIMIBI class (signal recognition, MinD and BioD) (52) of which several members have been identified to function as accessory proteins for metalloenzyme maturation. In some cases a single metal ion such as Ni 2ϩ or Zn 2ϩ bridges the homodimeric accessory protein for maturation of NiFe hydrogenase (HypB), urease (UreG), and carbon monoxide dehydrogenase (CooC) (53)(54)(55). In these accessory proteins the metal ion is postulated to be bound in a labile fashion and can be transferred to target proteins. This scenario may resemble the proposed scaffold function of Cfd1-Nbp35. These maturation systems clearly differ from the nitrogenase Fe-protein NifH, which maintains its bridging [4Fe-4S] cluster during its functional tasks, i.e. during the well known ATP hydrolysis-powered electron donation to the nitrogenase MoFe protein and during the more recently characterized biosynthetic function in reductive coupling of two [4Fe-4S] clusters for P cluster biosynthesis (56). Our study provides initial insight into the potential role of nucleotide binding/hydrolysis for Cfd1-Nbp35 function. We found that both proteins carrying mutations in the nucleotide binding motif could assemble a Fe-S cluster. However, for Fe-S cluster loading a functional nucleotide binding motif was required in trans. Apparently, nucleotide binding and/or hydrolysis is part of the mechanism of Cfd1-Nbp35-assisted Fe-S cluster assembly. We note that this mechanism differs from the de novo assembly of Fe-S clusters on Isu1, which does not require ATP hydrolysis by the Hsp70 chaperone Ssq1 (10,57). Rather, the ATP-dependent function of Ssq1 (or bacterial Hsc66) is needed for release of the Fe-S cluster from the scaffold protein (12). We cannot exclude, at this stage of our analyses, an additional requirement for nucleotide binding/hydrolysis during Fe-S cluster transfer to target apoproteins, as this step could not yet be assessed by our 55 Fe incorporation experiments.
Although our studies provide novel insights into the roles of two important functional motifs of Cfd1 and Nbp35 in Fe-S cluster assembly, it remains to be established by which biochemical reactions the Fe-S clusters are synthesized and what the individual roles of Cfd1 and Nbp35, both essential proteins in yeast, may be. From the involvement of the electron transfer chain Tah18-Dre2 in cytosolic Fe-S protein biogenesis (23) it seems possible that redox changes of the nascent Fe-S clusters or their synthesis intermediates may occur during these steps. Moreover, the dynamics of Fe-S cluster loading and transfer on the Cfd1-Nbp35 hetero-tetramer needs to be unraveled. Partial or even full dissociation of the Cfd1-Nbp35 complex may occur upon dissociation of the labile, bridging Fe-S cluster and transfer to target Fe-S proteins via other components of the CIA