Binding of the Chaperone Jac1 Protein and Cysteine Desulfurase Nfs1 to the Iron-Sulfur Cluster Scaffold Isu Protein Is Mutually Exclusive*

Background: Little is known regarding the dynamics of the interaction of proteins with the Fe/S cluster scaffold Isu1. Results: Three conserved Isu1 residues are critical for interaction with cysteine desulfurase Nfs1 and J-protein cochaperone Jac1, required for cluster assembly and transfer, respectively. Conclusion: Jac1 and Nfs1 binding to Isu1 are mutually exclusive. Significance: Mutual exclusivity suggests a point of regulation of the cluster assembly/transfer cycle. Biogenesis of mitochondrial iron-sulfur (Fe/S) cluster proteins requires the interaction of multiple proteins with the highly conserved 14-kDa scaffold protein Isu, on which clusters are built prior to their transfer to recipient proteins. For example, the assembly process requires the cysteine desulfurase Nfs1, which serves as the sulfur donor for cluster assembly. The transfer process requires Jac1, a J-protein Hsp70 cochaperone. We recently identified three residues on the surface of Jac1 that form a hydrophobic patch critical for interaction with Isu. The results of molecular modeling of the Isu1-Jac1 interaction, which was guided by these experimental data and structural/biophysical information available for bacterial homologs, predicted the importance of three hydrophobic residues forming a patch on the surface of Isu1 for interaction with Jac1. Using Isu variants having alterations in residues that form the hydrophobic patch on the surface of Isu, this prediction was experimentally validated by in vitro binding assays. In addition, Nfs1 was found to require the same hydrophobic residues of Isu for binding, as does Jac1, suggesting that Jac1 and Nfs1 binding is mutually exclusive. In support of this conclusion, Jac1 and Nfs1 compete for binding to Isu. Evolutionary analysis revealed that residues involved in these interactions are conserved and that they are critical residues for the biogenesis of Fe/S cluster protein in vivo. We propose that competition between Jac1 and Nfs1 for Isu binding plays an important role in transitioning the Fe/S cluster biogenesis machinery from the cluster assembly step to the Hsp70-mediated transfer of the Fe/S cluster to recipient proteins.

Iron-sulfur (Fe/S) clusters, ancient prosthetic groups found in all three domains of life, are present in a variety of proteins that function in essential cellular processes ranging from metabolism to the sensing of stress. In eukaryotic cells, mitochondria, which inherited their Fe/S cluster biogenesis system from bacterial ancestors, play a central role in maturation of cellular Fe/S cluster-containing proteins (1). Because the Saccharomyces cerevisiae Fe/S cluster biogenesis (ISC) system is the focus of this report, we use the gene/protein designation for this organism throughout. In the related bacterial and mitochondrial systems, the highly conserved small scaffold protein Isu serves as a platform on which a Fe/S cluster is assembled de novo prior to transfer to recipient proteins (1). In S. cerevisiae, Isu is encoded by a paralogous, functionally exchangeable gene pair, ISU1 and ISU2. Isu1 plays the major functional role because of its higher level of expression (2,3). Both the assembly and the transfer steps require the interaction of Isu with other proteins.
Several lines of evidence from bacterial and mitochondrial systems indicate that a "Fe/S cluster assembly complex" constitutes a functional and structural unit responsible for de novo synthesis of a cluster on the scaffold (4 -8). Most relevant to this report, Nfs1, a 51-kDa cysteine desulfurase, donates the sulfur needed for Fe/S cluster synthesis from cysteine, transferring it to Isu. In S. cerevisiae and other eukaryotes, Nfs1 is in a complex with the small accessory protein Isd11 (referred to as Nfs1(Isd11) throughout). Isd11 is proposed to both stabilize the Nfs1 protein and regulate its catalytic activity (9 -11). The yeast frataxin homolog Yfh1, which in humans is associated with Friedreich's ataxia, a neurological disease characterized by impairment of Fe/S cluster biogenesis and iron metabolism (12)(13)(14), is also part of the assembly complex. The function of Yfh1, which interacts with both Isu and Nfs1(Isd11), may be to serve as an iron donor and/or a regulator of cysteine desulfurase activity (4,15,16).
In both bacterial and mitochondrial ISC systems the Fe/S cluster transfer step is mediated by the J-protein-Hsp70 molecular chaperone system (17)(18)(19). A conserved specialized J-protein, called Jac1 in S. cerevisiae, is present in all eukaryotes and proteobacteria. As is typical in cases in which a J-protein also interacts with a client protein, the interaction of Jac1 with Isu is key, serving to target Hsp70 to its binding site on Isu (20 -22). It is thought that conformational changes of the cluster-containing scaffold induced upon Hsp70 interaction triggers the release of the Fe/S cluster from the scaffold and its transfer onto a recipient protein (23). Strikingly, although most J-proteins that interact with client proteins display a rather broad specificity, interacting with a wide array of client proteins, the interaction of Jac1 with the client is very specific. Isu is its only known client (19,20). In both the bacterial and mitochondrial systems, the C-terminal domain of Jac1 is directly responsible for Isu binding, with three hydrophobic residues playing a critical role in the interaction of Jac1 with Isu (22,24,25). Substitution of these residues with alanine sharply reduces the interaction of Jac1 with Isu in vitro and severely compromises both cell growth and the activity of the Fe/S cluster containing enzymes in vivo.
Not surprisingly, considering its central role in both the assembly and the transfer of Fe/S clusters, the Isu scaffold interacts with multiple other components of the Fe/S cluster biogenesis system (1). However, little information is available regarding the nature of these interactions and the functional consequences of their disruption. To begin to address these issues, we initiated a structure/function analysis. We identified three surface-exposed hydrophobic residues of Isu1 critical for its interaction with Jac1. These three residues were also critical for the interaction of Isu1 with the cysteine desulfurase Nfs1. Consistent with this dual role, Jac1 and Nfs1 competed with each other for Isu1 binding in vitro. On the basis of our experimental findings and the evolutionary conservation of residues involved in these protein-protein interactions, we hypothesize that the mutual exclusivity of these interactions plays a functional role in the transition between Fe/S cluster assembly and Fe/S cluster transfer steps in the biogenesis of Fe/S proteins.

EXPERIMENTAL PROCEDURES
Yeast Strains and Plasmids, Media, and Chemicals-All strains were of the S. cerevisiae W303 background. The double mutant having deletions of both ISU1 and ISU2 (26) is referred to as isu-⌬ throughout. To assess the level of the Nfs1 or Isu1 variants themselves or the activity of the Fe/S cluster containing enzymes, strains having the relevant WT gene at the normal chromosomal location under the control of the glucose-repressible GAL1-10 promoter were used (27). ISU1 and NFS1 mutants were generated in pRS314-ISU1 and pRS316-NFS1 using the Stratagene QuikChange protocol (26), as were all mutants in Escherichia coli expression vectors. JAC1 strains and plasmids have been described previously (22). Yeast was grown on YPD (1% yeast extract, 2% peptone and 2% glucose) or syn-thetic medium as described (28). All chemicals, unless stated otherwise, were purchased from Sigma.
Recombinant Jac1 His WT and mutant proteins were also purified as described previously (20), except E. coli strain C41(DE3) was used for expression. Recombinant Isu1-GST fusions were purified as described (22). In all cases, protein concentrations, determined using the Bradford (Bio-Rad) assay with bovine serum albumin as a standard, are expressed as the concentration of monomers.
Pull-down Assay-Titration pull-down experiments were performed by incubating the indicated concentrations of Jac1 His or Nfs1(Isd11) His with 2.5 M Isu1-GST in 150 l of PD buffer (40 mM Hepes-KOH (pH 7.5), 5% (v/v) glycerol, 100 mM KCl, 1 mM dithiothreitol, 10 mM MgCl 2 , and 1 mM ATP) for 30 min at 25°C to allow complex formation. Reduced glutathioneimmobilized agarose beads were pre-equilibrated with 0.1% bovine serum albumin, 0.1% Triton X-100, and 10% (v/v) glycerol in PD buffer. 40 l of beads (ϳ20-l bead volume) were added to each reaction and incubated at 4°C for 1 h with rotation. The beads were washed one time with 500 l and then three times with 200 l of PD buffer with 0.1% Triton X-100. Proteins bound to the beads were incubated with 2-fold-concentrated Laemmli sample buffer (20 l) for 10 min at 90°C, and 15-l aliquots were loaded on SDS-PAGE and visualized by Coomassie staining.
Cysteine Desulfurase Enzymatic Activity-The enzymatic activity of Nfs1(Isd11) was measured as sulfide production using cysteine as the substrate according to Ref. 29. In the standard assay, 0.5 M complex was incubated in 220 l of CD buffer (20 mM Tris-HCl (pH 8.0), 200 mM sucrose, 50 mM NaCl, and 6 mM dithiothreitol) supplemented with 10 M pyridoxal phosphate. The reaction was initiated by the addition of 0.5 mM L-cysteine. Following an incubation of 15 min at 25°C, the reaction was terminated by the addition of 0.1 ml of 20 mM N,N-dimethyl-p-phenylenediamine sulfate in 7.2 N HCl and 0.1 ml of 30 mM FeCl 3 in 1.2 N HCl. After further incubation in the dark for 20 min, the absorption of methylene blue was measured at 667 nm, and the sulfide concentration (nmol sulfide s Ϫ2 per min per mg protein) was calculated on the basis of an Na 2 S standard curve.
Mitochondrial Enzyme Activities-Activities of the respiratory enzymes were measured using mitochondria lysates as described previously in Ref. 30. Mitochondrial lysates were assayed for the activities of Fe/S cluster enzymes (succinate dehydrogenase and aconitase) and one non-Fe/S cluster protein (malate dehydrogenase) used here as a negative control. Succinate dehydrogenase activity was measured by using succinate as a substrate as described in Ref. 31. Aconitase activity was measured by monitoring the decrease in absorbance of the substrate isocitrate at 235 nm as described in Ref. 31. Malate dehydrogenase activity was measured using oxaloacetate as a substrate and by monitoring the decrease in absorbance of NADH at 340 nm as described in Ref. 31. Data were normalized to the protein content of the mitochondrial samples.
Levels of Mitochondrial Proteins-To quantify levels of Isu1 or Nfs1 variants, whole cell lysates were prepared by alkaline lysis (32) from 1 ml (A 600 ϭ 1.0) of cell culture. The cell pellet was washed once with 0.5 ml of 10 mM Tris/HCl and 1 mM EDTA (pH 8.0) and then resuspended in 0.5 ml of cold H 2 O. Cells were lysed by 10-min incubation after addition of 75 l of freshly prepared 1.85 M NaOH/7.4% 2-mercaptoethanol/10 mM PMSF. To precipitate protein, the mixture was incubated for 10 min on ice after addition of 575 l of 50% trichloroacetic acid. After centrifugation, pellets were washed twice with 1 ml of icecold acetone prior to drying. Proteins were resuspended by incubation at 95°C for 10 min after addition of 100 l of Laemmli sample buffer. Insoluble material was removed by centrifugation, and proteins in the supernatant were separated in SDS-PAGE gels. The resolved proteins were transferred electrophoretically to nitrocellulose. Isu1 or Nfs1 variants were detected by enhanced chemiluminescence (33) using anti-Nfs1 or anti-Isu polyclonal antiserum.
Prediction of Protein Structures-The crystallographic structure of E. coli IscU from the IscS-IscU complex (PDB code 3LVL (7)) was selected by GenSilico Metaserver (34) as the best scored template for homology modeling of S. cerevisiae Isu1. A structure model was obtained using MODELLER (35) on the basis of alignments of the target protein sequence to the template structure prepared using MUSCLE (36). With the Isu1 structure model as a ligand and the crystallographic structure of S. cerevisiae Jac1 (PDB code 3UO3 (22)) as a receptor the computational docking procedure was performed using the ZDOCK server (37) with exclusion of the Jac1 J-domain from potential interaction. From the results obtained, the bestscored model of the Jac1-Isu1 complex was chosen for further optimization. Combination of manual and computational structural refinement was applied using DeepView-Swiss-Pdb-Viewer (38) and Gaia (39), and short discrete molecular dynamics simulations were performed using Chiron (40). Of the optimized variants, the best-scored using FireDock (41) was chosen as the final model of the Jac1-Isu1 complex. Homology modeling of the S. cerevisiae Nfs1 structure was performed in the same way as Isu1, and the IscS from IscS-IscU (PDB code 3LVL (7)) was chosen as a structural template. Both Nfs1 and Isu1 models were overlaid with IscS-IscU crystallographic structure and, after an optimizing procedure the same as for the Jac1-Isu1 complex, a final model of the Nfs1-Isu1 complex was chosen as the best-scored using FireDock (41). Protein structure visualizations were prepared using the PyMOL Molecular Graphics System (version 1.5.0.4, Schrödinger, LLC).
Evolutionary Analysis-The eukaryotic orthologs of Jac1, Nfs1, and Isu1 were obtained with Basic Local Alignment Search Tool searches (42) performed against the available protein data of the individual species, with sequences of the S. cerevisiae ISU1, NFS1, and JAC1 genes used as queries. The bacterial orthologs were identified by retrieving proteobacterial sequences from the InterPro database entries IPR011339 (Isu1), IPR010240 (Nfs1), and IPR004640 (Jac1). The retrieved sequences were cross-filtered to include only sequences from species for which orthologs of all three genes could be identified. The obtained sequence datasets were aligned using MAFFT v7.023b with default options (43). Sequences for which it was impossible to determine with confidence positions homologous to the specific positions studied in this work were removed. See Supplemental Table S1 for complete data.

RESULTS AND DISCUSSION
Isu1 Residues Leu 63 , Val 72 , and Phe 94 Are Critical for Jac1-Isu1 Interaction-Previously (21,22), we defined a binding interface consisting of eight hydrophobic and charged residues on the surface of the C-terminal domain of Jac1 involved in interaction with Isu1 (Fig. 1A). As a first step in defining the residues of Isu1 involved in interaction with Jac1, we performed in silico protein-protein docking simulations using the Jac1 structure (PDB code 3UO3 (22)) and a homology model of the Isu1 structure on the basis of the crystal structure of E. coli IscU (PDB code 3LVL (7)) because the structure of yeast Isu has not been determined. We chose ZDOCK predictions having the highest score for optimization to obtain the model of the Jac1-Isu1 complex presented in Fig. 1A. In this model, Jac1 hydrophobic and negatively charged residues interact with residues, hydrophobic and positively charged, respectively, on the surface of Isu1. More specifically, the hydrophobic region of Jac1 composed of Leu 105 , Leu 109 , and Tyr 163 interacts with three hydrophobic residues of Isu1, Leu 63 , Val 72 , and Phe 94 , whereas the charged region of Jac1 (Asp 110 , Asp 113 , and Glu 114 ) interacts with a charged region of Isu1 encompassing residues Lys 54 , Lys 55 , and Arg 74 . Strikingly, this interface is consistent with experimental data published previously for both yeast and bacterial systems (22,24,25,44), although obtained independently. In particular, the hydrophobic residues (Leu 105 , Leu 109 , and Tyr 163 ) of Jac1 predicted by the modeling are the same as those found to be of critical importance for interaction with Isu1 in in vitro studies of engineered variants (22). In addition, the Isu1interacting residues obtained by modeling are among the residues of IscU, the E. coli Isu1 ortholog, for which chemical shift perturbations were observed by NMR spectroscopy upon addition of HscB (44), the Jac1 E. coli ortholog.
To test experimentally whether residues Leu 63 , Val 72 , and Phe 94 of Isu1, predicted to contact these three residues, are important for interaction with Jac1, we constructed two ISU1 mutants, substituting the codons for these three residues for those encoding either alanine or serine to generate variants Isu1 LVF/AAA and Isu1 LVF/SSS , respectively. Serine alterations were chosen in addition to the more typical alanine alterations because, unlike Jac1, in which the interacting hydrophobic residues were partially beneath the protein surface (Fig. 1A), Leu 63 , Val 72 , and Phe 94 of Isu1 are all highly exposed. We reasoned that replacement by alanines might have a lesser effect, compared with replacement of interacting residues on Jac1, on the size of the hydrophobic surface available for interaction (Fig.  1A). To test the ability of Isu1 LVF/AAA and Isu1 LVF/SSS to bind Jac1, we used a fusion between Isu1 and GST in a "pull-down assay" using purified proteins, as described previously (22). Increasing concentrations of Jac1 were incubated with a fixed concentration of Isu1-GST to allow complex formation. Glutathione resin was then used to pull down Isu1-GST and any Jac1 bound to it. We observed significantly less binding of Isu1 LVF/AAA -GST to Jac1 compared with WT Isu1-GST, indicating that the Leu 63 , Val 72 , and Phe 94 residues indeed contribute significantly to Jac1 binding (Fig. 1B). Consistent with our prediction, the effect of the serine substitutions on the Jac1-Isu1 interaction was more dramatic than that of alanine substitutions. No complex between Jac1 and Isu1 LVF/SSS was detected because the amount of the Jac1 pulled down by the Isu1 LVF/SSS -GST was indistinguishable from the background level (Fig. 1B). From these results, we concluded that three hydrophobic residues, Leu 63 , Val 72 , and Phe 94 , on the surface of Isu1 play a critical role in the formation of the Isu1-Jac1 complex in vitro.
Next, we wanted to assess the in vivo effects of the LVF/AAA and LVF/SSS alterations of Isu1, which have moderate and severe affects on Jac1 interaction, respectively. To this end, we transformed an isu-⌬ strain harboring a centromeric plasmid having a WT copy of the ISU1 gene and the URA3 marker with a second plasmid having a different selectable marker and carrying either the isu1 LVF/AAA or isu1 LVF/SSS gene. Cells were then plated on medium containing 5-fluorootic acid. Because only those cells having lost the plasmid containing the URA3 gene, and, therefore, the WT copy of ISU1, can grow on such a medium, the growth phenotype of cells expressing only an Isu1 variant can be assessed. Neither isu1 LVF/AAA nor isu1 LVF/SSS 5-fluorootic acid-resistant cells were recovered, indicating that neither variant can support growth (Fig. 1C). To ensure that the null phenotypes were due to altered protein function, not low expression, we used another isu-⌬ strain in which WT Isu1 expression was driven by the GAL-10 promoter and, thus, repressed upon glucose addition. After transformation of these Strains were plated on glucose-minimal medium containing 5-fluorootic acid, which selects for cells having lost the plasmid containing the URA3 marker, and incubated at 30°C for 3 days. D, lysates of GAL-ISU1:isu2-⌬ cells transformed with either a plasmid lacking an insert (-) or a WT copy of ISU1, isu1 LVF/AAA , isu1 LFV/SSS , under the control of the native ISU1 promoter, were prepared 17 h after transfer from galactose-to glucose-containing medium and separated by electrophoresis. Immunoblots were probed with antibodies specific to Isu1 and actin, a loading control.

Overlapping Binding of Nfs1 and Jac1 to Isu1
OCTOBER 4, 2013 • VOLUME 288 • NUMBER 40 JOURNAL OF BIOLOGICAL CHEMISTRY 29137 cells with a plasmid carrying a mutant ISU1 allele under the control of the native ISU1 promoter, cultures were shifted from galactose-to glucose-based medium. WT Isu1 was depleted below the level of immunodetection, whereas the levels of both Isu1 LVF/AAA and Isu1 LFV/SSS were similar to that of Isu1 in a WT strain (Fig. 1D). We conclude that, despite normal expression levels, neither Isu1 LVF/AAA nor Isu1 LVF/SSS are able to support cell growth.
Isu1 Residues Involved in Jac1 Interaction Are Important for Binding of Cysteine Desulfurase Nfs1-The inability of Isu1 LVF/AAA , which retained substantial affinity for Jac1, to support growth was surprising. Previously, we described a Jac1 variant, Jac1 Y163/A , having an alteration on its Isu1 interaction surface that resulted in a similar reduction in Jac1-Isu1 interaction as that observed for Isu1 LVF/AAA . However, Jac1 Y163/A was able to support robust growth (22). We decided to reevaluate Jac1 Y163/A (Fig. 1,  B and C). We also included Jac1 LLY/AAA in our analysis, which has alanine replacements of Leu 105 , Leu 109 , and Tyr 163 , the three hydrophobic residues found to be critically important for interaction with Isu1 in our earlier analysis (22). As expected, we found that Jac1 LLY/AAA had negligible affinity for Isu1 (Fig.  1B) and only supported very slow growth (C). This contrast between growth phenotypes and Jac1-Isu1 affinities suggested that residues Leu 63 , Val 72 , and Phe 94 of Isu1 may have a function(s) in addition to serving as an interface for interaction with Jac1. Inspection of structural data that recently became available for the bacterial orthologs of Isu and Nfs1 (IscU and IscS, respectively) is consistent with the idea that the Nfs1-Isu binding interface includes the Isu1 residues Leu 63 , Val 72 , and Phe 94 (7,8,45). To test whether these residues involved in the Jac1 interaction play a role in Nfs1 binding, we again used the Isu1-GST pull-down assay. When a fixed amount of Isu1-GST was incubated with increasing concentrations of purified Nfs1(Isd11), we observed that the amount of Nfs1 pulled down with Isu1-GST was concentration-dependent, with saturation reached at ϳ10 M Nfs1(Isd11) and half-maximal saturation at ϳ1.5 M Nfs1(Isd11) (Fig. 2). Thus, the affinity of Isu1 for Nfs1(Isd11) was similar to that observed for Jac1 (half-maximal saturation at ϳ 1 M concentration, Fig. 1B).
To test whether residues Leu 63 , Val 72 , and Phe 94 of Isu1 are indeed involved in the interaction with Nfs1(Isd11), we replaced these residues, either individually or in combination, with alanine. For Isu1 L63/A -GST and Isu1 V72/A -GST, we observed a substantial reduction of Nfs1(Isd11) binding in comparison to the Isu1-GST WT control (Fig. 2B). The effect of the Phe 94 /Ala replacement was less dramatic, with binding reduced by ϳ25%. The ability of Isu1 LVF/AAA -GST having the triple alanine substitution to bind Nfs1 (Isd11) was greatly reduced. Interaction of the LVF/SSS variant was less than 10% of the WT control, even at the highest concentration of Nfs1(Isd11) used in this experiment (Fig. 2B). These data indicate that the three residues of Isu1, Leu 63 , Val 72 , and Phe 94 , are critical for binding of both Jac1 and Nfs1(Isd11). Thus, a plausible explanation for the dramatic difference in phenotype of the jac1 and isu1 mutants (i.e. jac Y/A and isu1 LVF/AAA ) resulting in partial disruption of the Jac1-Isu1 interaction is that the isu mutants have a more severe effect on Nfs1 binding.
To further test the idea that the same residues of Isu interact with Jac1 and Nfs1, we decided to obtain a variant of Nfs1 defective in interaction with Isu1. We took advantage of structural information published previously about the complex of the bacterial orthologs (7,8) to model the Nfs1-Isu1 complex. Three hydrophobic residues in the C-terminal region of Nfs1 (Pro 478 , Leu 479 , and Met 482 ) were candidates for Isu1-interacting residues (Fig. 3A). Because alteration of a proline often leads to disruption of structural integrity, we choose to construct Nfs1 variants having alanine substituted at positions Leu 479 or Met 482 as well as in combination. We tested the ability of the purified Nfs1(Isd11) variants to interact with Isu1 using the Isu1-GST pull-down assay. The binding ability of both Nfs1 L479/A (Isd11) and Nfs1 M482/A (Isd11) variants was ϳ50% lower than that of WT protein, whereas binding of the variant having both alterations, Nfs1 LM/AA (Isd11), was about 80% lower (Fig. 3B). We conclude that residues Leu 479 and Met 482 of Nfs1 are critical for interaction with Isu1. This conclusion is in accord with data indicating the importance of the C-terminal segment of both Nfs1 (46) and its E. coli ortholog (7) for interaction with the scaffold. We note that it is likely that residues in addition to Leu 479 and Met 482 play important roles in the interaction of Nfs1 with Isu1. To assess the biological importance of the interaction of Nfs1 with Isu1 mediated by Leu 479 and Met 482 , we carried out in vivo experiments. We found that cells expressing Nfs1 LM/AA were not viable (Fig. 3C) even though protein was expressed at the WT level. However, the cysteine desulfurase activity of Nfs1 LM/AA was similar to that of the wild type (Fig. 3D), supporting the idea that these substitutions did not globally affect the structural properties of Nfs1. To ensure that the growth defects of the ISU1 and NFS1 mutants we observed were due to effects on Fe/S cluster biogenesis, we also tested the activity of two cluster-containing mitochondrial enzymes, aconitase and succinate dehydrogenase (SDH) 5 . As expected, the activity of aconitase and SDH was severely affected after depletion of WT Isu1 from cells expressing Isu1 LVF/AAA (Fig. 3E) or after depletion of WT Nfs1 from cells expressing Nfs1 LM/AA (F), consistent with the idea that direct interaction of Isu1 with Nfs1 is critical for the biogenesis of Fe/S cluster proteins.
Competition between Nfs1 and Jac1 for Isu1 Binding-Because the same three residues (Leu 63 , Val 72 , and Phe 94 ) are important for the interaction of Isu1 with both Jac1 and Nfs1, we predicted that these two proteins directly compete for Isu1 binding (Fig. 4). To test this idea, we set up a biochemical competition assay on the basis of the Isu1-GST pull-down technique that we used to study the individual protein-protein interactions. First, we incubated a fixed amount of Isu1-GST (2.5 M) with a fixed 5 M concentration of Nfs1(Isd11), allowing the formation of the Isu1-GST-Nfs1(Isd11) complex. Next, increasing concentrations of Jac1 were added to the reaction 5 The abbreviation used is: SDH, succinate dehydrogenase. , WT or variants, as indicated, were mixed with Nfs1(Isd11) at the indicated concentrations to allow complex formation. Glutathione resin was added to pull down the complex. Proteins were separated by electrophoresis, visualized by staining, and quantitated by densitometry. Values were plotted in GraphPad Prism using a 1:1 binding hyperbola to fit data and plotted as relative units (r.u.), with maximal binding of WT Nfs1 given a value of 1. L479, Nfs1 L479/A ; M482, Nfs1 M482/A ; LM, Nfs1 L479,M482/AA . C, top panel, nfs1-⌬ cells harboring an URA3-marked plasmid containing the WT NFS1 (WT) and a second plasmid harboring either WT NFS1 or nfs1 L479,M482/AA (LM) were plated on glucose-minimal medium containing 5-fluorootic acid, which selects for cells having lost the plasmid containing the URA3 marker. The plate was incubated at 30°C for 3 days. Bottom panel, lysates of GAL-NFS1 cells transformed with plasmids having no insert (-) or harboring either a WT copy of NFS1 (WT) or nfs1 L479,M482/AA under the control of the native NFS1 promoter were prepared 22 h after transfer from galactoseto glucose-based medium and separated by SDS-PAGE. Immunoblots were probed with antibodies specific to Nfs1 and porin, a loading control. D, cysteine desulfurase activity of purified WT and the Leu 479 Met 482 /AA variant (LM) Nfs1(Isd11) was measured. E, aconitase activity (Aco) and succinate dehydrogenase activity were measured in lysates of mitochondria isolated from GAL-ISU1 isu2-⌬ cells harboring plasmid-borne copies of ISU1 (WT), Isu1 L63V72F94/AAA isu1 LVF/AAA , or vector without insert (-), grown for 17 h after transfer from galactose-to glucose-containing medium. As a standard, the enzymatic activity of non-Fe/S cluster-containing protein malate dehydrogenase (MDH) was measured. The ratio of activities of aconitase or SDH and malate dehydrogenase was calculated and expressed as a percentage of the WT control. Bars represent average values for three measurements, with presented error bars as S.D. F, aconitase activity and SDH activity were measured in lysates of mitochondria isolated from GAL-NFS1 cells harboring plasmid-borne copies of WT NFS1, nfs1LM/AA, or vector without insert, as indicated, grown for 40 h in glucose-containing medium. Enzymatic activities were measured and plotted as described in E. Overlapping Binding of Nfs1 and Jac1 to Isu1 OCTOBER 4, 2013 • VOLUME 288 • NUMBER 40 mixtures, followed by pull-down with glutathione resin. With an increasing concentration of Jac1, we observed a decreasing amount of Nfs1 associated with Isu1-GST so that, at 10 M Jac1, the amount of Nfs1 in complex with Isu1-GST was reduced to Ͻ 50% of the initial value (Fig. 4A). When Jac1 WT was replaced by the Jac1 LLY/AAA variant, which is strongly defective in Isu1 binding, no displacement of Nfs1(Isd11) associated with Isu1-GST was observed over a wide range of Jac1 LLY/AAA concentrations. These results indicate that Jac1 and Nfs1 binding is mutually exclusive.
To verify this idea, we performed a reverse competition experiment. First, we incubated a fixed amount of Isu1-GST (2.5 M) with a fixed amount of Jac1 (5 M), allowing the formation of the Isu1-GST:Jac1 complex (Fig. 4B). Subsequently, either WT Nfs1(Isd11) or Nfs1 L479,M482/AA (Isd11) was added to the reaction mixture. With increasing concentrations of WT Nfs1(Isd11), we observed decreasing amounts of Jac1 associated with Isu1-GST so that, at 5 M Nfs1(Isd11), the amount of Jac1 in complex with Isu1-GST was reduced to ϳ50% of the initial value (Fig. 3B). However, no reduction in the amount of bound Jac1 was observed when Nfs1 L479,M482/AA (Isd11) was added as a competitor, even when added at a concentration of 20 M (Fig. 4B). This result is consistent with the idea that Leu 479 and Met 482 are indeed important for Nfs1-Isu1 interaction. Overall, we conclude that Nfs1 and Jac1 compete for overlapping binding sites on the surface of Isu1 that contain the three hydrophobic residues Leu 63 , Val 72 , and Phe 94 .
Residues Involved in Jac1-Nfs1 Interactions with Isu1 Are Evolutionary Conserved-The experiments described above were obtained using S. cerevisiae as the model system. To place our findings in an evolutionary context, we compared residues homologous to those involved in Jac1-Nfs1 interactions with Isu1 across the phylogeny. To this end, we identified orthologs of Jac1, Nfs1, and Isu1 from fully sequenced genomes of 84 eukaryotic species (62 fungal and 22 other eukaryotic species, including most model organisms) and 390 proteobacteria species (supplemental Table S1). Our analysis revealed that sites involved in Nfs1 and Jac1 interactions with Isu1 are indeed evolutionarily conserved because they are either invariant across the phylogeny or they are occupied by highly similar amino acid residues ( Fig. 5 and supplemental Table S1). For example, in only a limited number of eukaryotic Nfs1 orthologs, Met 482 is replaced by leucine, and Tyr 163 of the Jac1 orthologs is replaced by phenylalanine.  Table S1 for the list of species). The percentage of species having given residues is indicated. FIGURE 6. Jac1 involvement in the transition from Fe/S assembly to transfer. A, when an Fe/S cluster is synthesized on the Isu1 scaffold via action of the assembly complex (left), Jac1 displaces Nfs1 (center) and Jac1 targets Isu1-Fe/S for mtHsp70 binding, facilitating cluster transfer to recipient apoprotein (right). B, homology model of Isu1 with highlighted residues involved in both Nfs1 and Jac1 binding (orange), Jac1 binding only (brown), and Nfs1 binding only (yellow).
Interestingly, we also observed a pattern of residue conservation consistent with the hypothesis that ␣-proteobacteria are closely related to ancestors of mitochondria because residues conserved in mitochondrial orthologs are identical to those of ␣-proteobacteria, such as Rickettsia (47), and different from those of other bacterial species. For example, Leu 63 and Phe 94 of Isu1 are shared among mitochondrial and ␣-proteobacteria orthologs, whereas orthologs from other proteobacteria species have mostly methionine and tyrosine at homologous positions. Overall, phylogenetic analysis revealed that the molecular mechanism of Nfs1-Jac1 interactions with Isu1 is evolutionary conserved and that mitochondrial proteins inherited residues involved in those interactions from their ␣-proteobacterial ancestor.

CONCLUSIONS
Several lines of evidence indicate that the cysteine desulfurase Nfs1 and the J-protein cochaperone Jac1 bind to overlapping sites on Isu1, each containing the hydrophobic residues Leu 63 , Val 72 , and Phe 94 . These highly conserved overlapping sites render their interactions with Isu1 mutually exclusive. A central issue in Fe/S cluster biogenesis via the ISC pathway is understanding the transition from the process of assembly of the Fe/S cluster on Isu to the process of transfer (Fig. 6A). It is well accepted that Hsp70 is critical for cluster transfer and that the Jac1-Isu1 interaction is critical for targeting Isu1 for Hsp70 binding (1,19,48). The results reported here, indicating the exclusive nature of the Jac1 and Nfs1 interaction with Isu, suggest that an ordered transition occurs from cluster assembly to cluster transfer.
What could be the driving force(s) behind such an ordered transition? Two factors seem most likely to play a role. First, the relative affinities of Nfs1 and Jac1 may differ for the apo-and holoforms of the scaffold (6,49). Higher affinity for the apoform compared with the holoform of Isu1 for Nfs1 (and vice versa for Jac1) would serve to drive the transition because of more efficient competition of Jac1 for binding to holoIsu upon dissociation of Nfs1. Secondly, it is possible that Jac1 plays a more active role. We note that the Nfs1 and Jac1 binding sites on Isu1 only overlap partially (22). Notably, on the basis of the available structural and biochemical data, the charged patch of Isu1 involved in the Jac1-Isu interaction does not appear to participate in Nfs1 interaction (Fig. 6B). Thus, it is possible that Jac1 interaction with Isu may be initiated even when Nfs1 is in contact with Isu1, allowing Jac1 to actively participate in a "kinetic switch" between cluster assembly and transfer. It should also be remembered that Yfh1, which interacts with both Isu and Nfs1 (4 -6, 14 -16), plays a critical role in efficient biogenesis of Fe/S clusters. Available structural and biochemical data suggest that those sites do not overlap with Nfs1-Isu1 interaction, but little information is available about its effects on the dynamics of interaction of other components with Isu1. Further work will be required to understand the dynamics of interactions between the cluster-bound and cluster-free forms of the scaffold and the interacting components involved in assembly and transfer.