Isd11p Protein Activates the Mitochondrial Cysteine Desulfurase Nfs1p Protein*

Background: In yeast, Nfs1p is the only known cysteine desulfurase in mitochondria. Results: Nfs1p is completely inactive on its own and requires an accessory protein, Isd11p, for activity. Conclusion: Binding of Isd11p may induce a conformational change in Nfs1p, thereby activating the enzyme. Significance: Isd11p-mediated activation of Nfs1p is critical for vital cellular and mitochondrial functions. Cysteine desulfurases perform pyridoxal phosphate (PLP)-dependent desulfuration of cysteine. The key steps of the enzymatic cycle include substrate binding to PLP, formation of a covalent persulfide intermediate at the active site cysteine, and transfer of sulfur to recipients for use in various metabolic pathways. In Saccharomyces cerevisiae, the cysteine desulfurase Nfs1p and an accessory protein, Isd11p, are found primarily in mitochondria, and both are essential for cell viability. Although cysteine desulfurases are conserved from bacteria to humans, Isd11p is found only in eukaryotes and not in prokaryotes. Here we show that Isd11p activates Nfs1p. The enzyme without Isd11p was inactive and did not form the [35S]persulfide intermediate from the substrate [35S]cysteine. Addition of Isd11p to inactive Nfs1p induced formation of the persulfide. Remarkably, in a two-step assay, [35S]cysteine could be bound to the inactive Nfs1p in a PLP-dependent manner, and the enzyme could be subsequently induced to form the persulfide by addition of Isd11p. A mutant form of Isd11p with the 15LYK17 motif changed to 15AAA17 was able to bind but failed to activate Nfs1p, thus separating these two functions of Isd11p. Finally, compared with Nfs1p with or without the bound Isd11p mutant, the Nfs1p·Isd11p complex was more resistant to inactivation by an alkylating agent. On the basis of these novel findings, we propose that interaction of Isd11p with Nfs1p activates the enzyme by inducing a conformational change, thereby promoting formation of the persulfide intermediate at the active site cysteine. Such a conformational change may protect the active site cysteine from alkylating agents.

the amino acid cysteine, is accomplished after reduction of sulfate to sulfide. The addition of sulfide converts O-acetyl serine to cysteine, which can then be used for generating important metabolites, including methionine, S-adenosylmethionine, coenzyme A, and glutathione. By contrast, an activated form of sulfur, called sulfane sulfur (S 0 ) or persulfide (R-S-SH), is derived from cysteine by the action of cysteine desulfurases. It is this form of sulfur that is exclusively utilized in the synthesis of cofactors such as iron-sulfur (Fe-S) clusters, biotin, lipoic acid, and thiamin, and also for the thio modification of tRNAs.
Cysteine desulfurases generate persulfide from cysteine and donate sulfur to designated recipients. These enzymes require PLP 2 as the cofactor and are conserved from bacteria to humans. The enzymatic mechanism of the prokaryotic cysteine desulfurase has been defined by the pioneering work of Dean and co-workers on Azotobacter vinelandii NifS (4,5). Briefly, the salient features of the enzymology include a conserved lysine forming a Schiff base with the PLP cofactor, formation of a substrate cysteine-PLP ketimine adduct, nucleophilic attack by the thiolate anion of the active site cysteine on the sulfhydryl group of the substrate cysteine-PLP adduct to cleave the C-S bond, and formation of enzyme-bound persulfide at the active site cysteine with concomitant release of alanine thus formed and regeneration of PLP for a new catalytic cycle (5)(6)(7). Recent studies have pointed to a second portion of the prokaryotic cysteine desulfurase reaction involving direct sulfur transfer from the persulfide at the active site to specific recipients. Mutations in the peptide loop containing the catalytic site are able to abrogate sulfur delivery for Fe-S cluster synthesis without affecting the corresponding process for thiolation of some tRNAs, demonstrating the specificity of this transfer step (8). Purified NifS exhibits high cysteine desulfurase activity and does not require any other interacting or accessory proteins for this activity. The enzyme is extremely sensitive to low concentrations of alkylating agents such as N-ethylmaleimide (NEM) because of inactivation of the active site cysteine (4,5).
The structures of Escherichia coli IscS (9) and Thermotoga maritima NifS (10) are similar. They form homodimers and belong to the ␣-family of PLP-dependent enzymes. Each monomer contains two domains: the larger harboring the PLP cofactor and the smaller containing the active site cysteine in the middle of a highly flexible loop. The PLP cofactor is situated at the base of a pocket near the surface of the protein, and the pocket forms the substrate-binding site. The loop with the active site cysteine is not well resolved in these structures. However, in the IscS crystals with partially ordered loop, the active site cysteine appears Ͼ17Å away from the PLP cofactor. This implies that large conformational changes in this region would be needed during catalysis (9). To form the persulfide, the active site cysteine would have to rotate toward the PLP site, whereas delivery of the persulfide might involve movement away from the substrate-binding site. No structures of eukaryotic cysteine desulfurases have been solved to date.
In eukaryotic cells such as the yeast Saccharomyces cerevisiae, the cysteine desulfurase is encoded by a single gene, NFS1. The Nfs1p protein is found primarily in mitochondria (11), although a small amount appears to function in extramitochondrial locations (12). Nfs1p is essential for cell viability. It is required for activities of Fe-S cluster proteins (11,13). Point mutations in the conserved lysine for PLP attachment or in the conserved active site cysteine result in cell inviability, suggesting that Nfs1p-mediated persulfide formation is critical for Fe-S cluster synthesis (11). This notion is consistent with the observation that a hypomorphic nfs1 mutant exhibits high cellular iron uptake and accumulates iron in mitochondria, similar to those observed in other yeast mutants that are deficient in mitochondrial Fe-S cluster assembly (11). Furthermore, Nfs1p appears to be required for thiolation of some of the tRNAs involved in mitochondrial protein synthesis (14). Interestingly, nfs1 mutants are capable of suppressing the lysine auxotrophy of cells lacking the Cu/Zn superoxide dismutase Sod1p (15). Like the yeast Nfs1p, human Nfs1 is also found primarily in mitochondria, with a small amount in the cytosol/nucleus, and knockdown of the human enzyme is lethal likely because of strongly impaired activities of both mitochondrial and cytosolic Fe-S cluster proteins (16,17).
In 2006, two groups identified Isd11p (approximately 11 kDa) on the basis of essentiality and mitochondrial localization in yeast (18,19). Although Isd11p is conserved in eukaryotes (20), no prokaryotic homolog has been found. This is remarkable for a component of the mitochondrial Fe-S cluster assembly machinery, because other components of this machinery have homologs in bacteria. Isd11p may, therefore, perform a unique function in mitochondria that is not needed in bacteria. Isd11p has been found to associate with Nfs1p in mitochondria, forming a complex (Nfs1p⅐Isd11p) of approximately 200 kDa. A temperature-sensitive allele or Isd11p-depleted cells show deficiencies of Fe-S cluster protein activities (18,19). Likewise, knockdown of the corresponding Isd11p homologs in Trypanosoma brucei (21) and HeLa (22) cells results in inactivation of Fe-S cluster-containing proteins. However, the function of Isd11p has not been determined, and consequently, the reason for Fe-S cluster deficiency in Isd11p-depleted mitochondria remains elusive. Iron-sulfur cluster assembly in mitochondria is a highly complex process requiring multiple proteins (23-25).
Here we used purified proteins to directly determine the role of Isd11p in Nfs1p activity. We demonstrate that Isd11p is required for the Nfs1p cysteine desulfurase activity. Binding of Isd11p likely induces a conformational change in the enzyme that allows persulfide formation at the active site cysteine.

EXPERIMENTAL PROCEDURES
Expression of Proteins in Bacteria-The ORF for the mature form of Nfs1p (lacking the amino-terminal 36 amino acids of the corresponding precursor protein) was amplified by PCR from a yeast genomic library using appropriate primers so that the resulting product was NdeI-ORF-XhoI. The ISD11 ORF was also amplified in a similar manner. The PCR products were digested with NdeI and XhoI, cloned into the same sites of pET21b (Novagen), and sequenced. This introduces a His 6 tag in frame at the C terminus of the proteins. The plasmid pT7-7/NifS (pDB551) for expression of A. vinelandii NifS was a gift from Dr. Dennis R. Dean (4). The ORF was modified by introduction of a 5Ј NdeI restriction site and a 3Ј His 6 tag followed by a stop codon and BamHI site. The resulting plasmid was called pT7-7/NifS-His 6 .
The plasmids pST39 and pET3aTr were obtained from Dr. Song Tan (26). The PCR product corresponding to NdeI-mature Nfs1p-His 6 -stop-BamHI was cloned into pET3aTr using NdeI and BamHI restriction sites. The XbaI-BamHI fragment of the resulting plasmid, now containing a 5Ј Shine-Dalgarno sequence, was moved to pST39 in the corresponding sites. For insertion of the ISD11 ORF, a number of steps were required. A stop codon and a BamHI site were introduced immediately after the ISD11 ORF in pET21b, and the NdeI-BamHI fragment was moved to pET3aTr. The SacI-Kpn1 fragment of the resulting plasmid was transferred to pST39, picking up the Shine Dalgarno binding site in the process. The HindIII fragment containing the ORF and flanking regions was moved to the pST39 version that already contained the sequence encoding Nfs1p-His 6 . The final construct (plasmid 20 -38) contained a T7 promoter driving a polycistronic mRNA for mature Nfs1p-His 6 and Isd11p, each with separate ribosome-binding sites. Using the QuikChange site-directed mutagenesis kit (Stratagene), the residues 15 LYK 17 in Isd11p were changed to 15 AAA 17 in pET21b/Isd11p-His 6 and also in plasmid 20 -38 containing both Nfs1p-His 6 and Isd11p. The mutations were confirmed by sequencing.
All proteins used for our studies described here were expressed in BL21 (DE3) Codon Plus cells (Stratagene). These cells, carrying different plasmids, were initially cultured in Lysogeny Broth supplemented with 100 g/ml ampicillin and 34 g/ml chloramphenicol to A 600 of approximately 0.6. Conditions for expression of proteins were optimized so that the majority of each protein was in soluble form (27,28). Variations included temperature and period of induction and also the concentration of the inducer isopropyl-1-thio-␤-D-galactopyranoside. Expression of Nfs1p-His 6 alone or together with Isd11p (with or without mutation in the 15 LYK 17 motif) was carried out in the presence of 0.5 mM isopropyl-1-thio-␤-D-galactopyranoside for 24 h at 20°C. Isd11p-His 6 by itself (with or without mutation) was expressed with 0.2 mM isopropyl-1-thio-␤-Dgalactopyranoside for 72 h at 25°C, and in this case, 0.5 M sorbitol was included during induction to further enhance the sol-ubility of the protein (28). A. vinelandii NifS-His 6 was induced with 0.5 mM isopropyl-1-thio-␤-D-galactopyranoside for 3 h at 37°C. For Nfs1p-His 6 alone, Nfs1p-His 6 together with Isd11p (Nfs1p⅐Isd11p), and A. vinelandii NifS, cells were harvested and washed with buffer A (50 mM Tris/HCl (pH 8.0), 0.15 M NaCl, 10% glycerol, 1 mM PMSF). Cells in buffer A were treated with lysozyme (50 g/ml) for 30 min on ice and then disrupted using a probe sonicator (Branson Sonifier 450) by six continuous 15 s bursts at 2 min intervals while keeping the samples on ice. Subsequent steps were carried out at 4°C. Cell lysates were centrifuged at 12,000 ϫ g for 30 min, and the supernatant fraction was incubated with Ni NTA agarose by end-over-end mixing for 3 h. After washing the resin with buffer A containing 10 mM imidazole, bound proteins were eluted with 0.4 M imidazole in buffer A. Purified proteins were stored in aliquots at Ϫ80°C until further use. Isd11p-His 6 was also purified and stored exactly the same way, but NaCl was left out of buffer A.
Sucrose Density Gradient Centrifugation-Mitochondria were purified from a wild-type S. cerevisiae strain D273-10B (ATCC 24657) (29). Mitochondria were suspended in buffer B (50 mM Tris/HCl (pH 7.5), 2 mM MgCl 2 , 2 mM DTT) containing 1 mM ATP and 1 mM NADH, and membranes were ruptured by mild sonication. Following addition of KCl (0.25 M final) to release membrane-associated Nfs1p and Isd11p, samples were centrifuged at 18,000 ϫ g for 15 min. The supernatant was loaded on a 5-16% sucrose gradient in buffer B containing 0.25 M KCl and centrifuged at 4°C for 22 h at 285,000 ϫ g (Beckman SW40 Ti rotor). Twenty fractions (approximately 0.62 ml each) were collected from the top of the centrifuge tube. Alternate fractions were analyzed by SDS-PAGE (with DTT present) followed by immunoblotting using anti-Nfs1p and anti-Isd11p antibodies. Bacterially expressed and purified Nfs1p⅐Isd11p complex was loaded on a separate sucrose gradient in the same buffer, and fractions were analyzed in the same manner. Likewise, protein standards were loaded on another gradient. In this case, fractions were analyzed by SDS-PAGE followed by Coomassie Blue staining to identify the peak fractions (30).
Cysteine Desulfurase Assays-A typical reaction mixture (50 l) contained proteins in buffer C (20 mM Hepes/KOH (pH 7.5), 0.6 M sorbitol, 0.15 M NaCl) containing 150 M PLP and 5 Ci [ 35 S]cysteine (0.1 M; 1075 Ci/mmol). Following incubation at 30°C for 10 min, reactions were stopped, and proteins were precipitated with the addition of an equal volume of icecold 20% TCA. Samples were centrifuged at 15,000 ϫ g for 30 min, and the pellets were analyzed by non-reducing SDS-PAGE followed by autoradiography. Imidazole did not interfere with these assays up to a final concentration of 40 mM. Variations of these assays included different amounts of the proteins, different concentrations of PLP and [ 35 S]cysteine, and different incubation temperatures and time periods. In some cases, proteins were treated with NEM and/or DTT prior to the assay. Proteins were then precipitated by ammonium sulfate (final 67% saturation) to remove free NEM/DTT. The protein pellets were solubilized with buffer A and assayed. These variations are outlined in the corresponding figures. Radiolabeled protein bands were quantitated using the Adobe Photoshop CS4 software.
The cysteine desulfurase activity was also measured by a spectrophotometric assay as follows. Reaction mixtures (350 l) in sealed Eppendorf tubes were prepared with various amounts of purified proteins in buffer D (50 mM Tris/HCl (pH 8.0), 1.5 mM L-cysteine, 5 mM DTT, 200 M PLP). Samples were incubated for 1 h at 37°C. The reaction was stopped by addition of 6% NaOH (50 l) and water (250 l). Sulfide released was measured as described (31). Briefly, 125 l of 0.1% DPD (N,Ndimethyl-p-phenylenediamine dihydrochloride in 5N HCl) and 50 l of 11.5 mM FeCl 3 were added, followed by incubation for 20 min at 37°C. Samples were centrifuged at 15,000 ϫ g for 5 min, and the absorbance of the supernatant was measured at 670 nm.

Bacterial Expression of Proteins and Characterization of the Nfs1p⅐Isd11p
Complex-The precursor form of the yeast mitochondrial Nfs1p contains 497 amino acids, including the 36amino acid-long targeting signal at the N terminus of the protein. Upon import into mitochondria, the targeting signal is proteolytically processed in two steps. The first 33 amino acids are removed by the mitochondrial processing peptidase and then another three amino acids by a newly identified peptidase, Icp55p (32). At the time of previous studies involving bacterial expression of Nfs1p, the exact length of the cleavable targeting sequence was unknown, and residue 68 was arbitrarily chosen as the N terminus of the mature protein (33,34). For our studies described here, we used the authentic mature form of Nfs1p (approximately 51 kDa) that lacks the first 36 amino acids of the corresponding precursor protein (approximately 55 kDa). Unlike in the case of Nfs1p, the mitochondrial targeting signal of Isd11p does not appear to be proteolytically removed upon import into mitochondria (18,19).
The mature form of Nfs1p, Isd11p, and A. vinelandii NifS, each with a C-terminal His 6 tag, were separately expressed in E. coli. Conditions were optimized so that the major portion (approximately 60 -80%) of the proteins remained soluble. All three proteins were purified to homogeneity by Ni-NTA affinity chromatography (Fig. 1A). The proteins Nfs1p-His 6 and Isd11p were also coexpressed in bacteria from a polycistronic vector. Both proteins were found in the soluble fraction. These proteins formed a complex (Nfs1p⅐Isd11p) because Isd11p with no His 6 tag was copurified with Nfs1p-His 6 by Ni-NTA chromatography (Fig. 1A, lane 2). The purified complex was yellow because of its association with the PLP cofactor.
The molecular nature and mass of the bacterially expressed Nfs1p⅐Isd11p complex were then compared with those of the mitochondrial Nfs1p⅐Isd11p complex by sucrose density gradient centrifugation. Fractions were analyzed by SDS-PAGE followed by immunoblotting using antibodies against Nfs1p and Isd11p. Mitochondrial Nfs1p and Isd11p were found to form a peak in fractions 9 -11 (Fig. 1B, panels 1 and 2), and neither Nfs1p nor Isd11p were detected elsewhere in the gradient. On the basis of the migration of standard proteins, the molecular mass of the mitochondrial complex was calculated to be approximately 200 kDa. These results are in good agreement with FPLC or Blue Native gel analysis of the mitochondrial complex (18,19). Interestingly, bacterially expressed and purified Nfs1p⅐Isd11p complex also formed a major peak in fractions 9 -11 (Fig. 1B, panels 3 and 4). An additional minor (approximately 10%) peak of Nfs1p observed for the bacterially expressed sample (Fig. 1B, panel 3, fraction 5) likely represents free Nfs1p-His 6 , which was not assembled with Isd11p but was copurified with the Nfs1p⅐Isd11p complex by Ni-NTA chromatography. No free Isd11p was detected (Fig. 1B, panel 4). These results suggest that the bacterially expressed and purified Nfs1p⅐Isd11p complex is similar to the mitochondrial complex, and, therefore, experiments with the purified complex are likely to mimic the mitochondrial scenario.
Purified Nfs1p⅐Isd11p Complex but Not Nfs1p by Itself Exhibits Cysteine Desulfurase Activity-The enzymatic cycle of cysteine desulfurase involves formation of a covalent persulfide at the active site cysteine. To assess persulfide formation, purified proteins were incubated with [ 35 S]cysteine and analyzed by non-reducing SDS-PAGE followed by autoradiography. For the Nfs1p⅐Isd11p complex, results show a protein concentrationdependent 35 S signal associated with Nfs1p ( Fig. 2A, lanes 1 and  3). Radiolabeled Nfs1p most likely represents [ 35 S]persulfide formed on Nfs1p of the complex. As expected, treatment with a reducing agent such as DTT completely released the persulfide sulfur, and consequently, no radiolabeled Nfs1p was detected ( Fig. 2A, lane 5). Isd11p, which was present in the starting complex and then separated by the denaturing conditions of the gel, was not radiolabeled. More importantly, Nfs1p alone, incubated with [ 35 S]cysteine prior to resolution on non-reducing SDS gel, was not radiolabeled even at the highest amount of the protein tested ( Fig. 2A, lanes 2 and 4). Thus, radiolabeled persulfide was formed on Nfs1p of the Nfs1p⅐Isd11p complex but not on Nfs1p alone, reflecting cysteine desulfurase activity of the complex and lack of activity of Nfs1p by itself.
Cysteine desulfurases require PLP as the cofactor. Therefore, to further characterize persulfide formation by the Nfs1p⅐Isd11p complex, the PLP requirement was assessed. As expected, the appearance of radiolabeled Nfs1p was greatly enhanced with increasing concentrations of added PLP, reaching a maximum at approximately 100 -500 M PLP (Fig. 2B). Interestingly, a small amount of radiolabeled Nfs1p was detected in the absence of added PLP. This is most likely due to the presence of PLP already bound to Nfs1p during expression in bacteria, as indicated by the yellow color of the purified Nfs1p⅐Isd11p complex. Likewise, [ 35 S]persulfide formation on Nfs1p by the Nfs1p⅐Isd11p complex was greatly stimulated with increasing concentrations of [ 35 S]cysteine (Fig. 2C). As expected, persulfide formation was temperature-dependent (Fig. 2D). However, the persulfide was detected within seconds of incubation, even on ice (Fig. 2E). These enzymatic properties of the Nfs1p⅐Isd11p complex are very similar to prokaryotic cysteine desulfurases (35).
The [ 35 S]persulfide formation in the assays described above represents a single round of activity. We therefore further tested the proteins using a spectrophotometric assay that measures continuous persulfide formation and release of persulfide sulfur as sulfide (31). In this case, proteins were incubated with unlabeled cysteine and DTT, and the quantity of released sulfide in solution was assessed after addition of N,N-dimethyl-pphenylenediamine dihydrochloride and ferric chloride. The Nfs1p⅐Isd11p complex was active, and the activity was concentration-dependent and linear (Fig. 2F). However, no activity of Nfs1p by itself or Isd11p was detected. Thus, Nfs1p alone is inactive, Isd11p is inactive, and only the Nfs1p⅐Isd11p complex is active by two different assays. The spectrophotometric assay required a minimum of 3-5 g of proteins. By contrast, [ 35 S]persulfide formation as described above can be easily detected with 50 ng of proteins. Because of the high sensitivity, all subsequent experiments were performed with [ 35 S]cysteine looking for Nfs1p-S-35 SH formation.
Isd11p Activates Nfs1p to Form the Persulfide Intermediate-Isd11p does not appear to be required for solubility and/or stability of Nfs1p because bacterially expressed Nfs1p by itself is soluble and the purified protein is stable with no detectable aggregation or degradation during storage. These observations allowed us to test the possibility that Isd11p is specifically required for the cysteine desulfurase activity of Nfs1p. For this purpose, Nfs1p was incubated with [ 35 S]cysteine in the absence or presence of added Isd11p and analyzed by non-reducing SDS-PAGE followed by autoradiography. Although Nfs1p was not radiolabeled in the absence of Isd11p (Fig. 3A, lane 2), addition of Isd11p resulted in strong radiolabeling (Fig. 3A, lane 3). This radiolabeling was due to [ 35 S]persulfide formation on the enzyme because the Nfs1p-associated 35 S signal was completely lost upon treatment with DTT (Fig. 3A, lane 5). Interestingly, the efficiency of [ 35 S]persulfide formation on Nfs1p with Isd11p added separately (Fig. 3A, lane 3) was comparable with that of bacterially coexpressed and preformed Nfs1p⅐Isd11p complex (lane 1). As expected, no radiolabeled band was detected in samples containing only Isd11p (Fig. 3A, lane 4). These results suggest that individually purified Nfs1p and Isd11p readily interact with each other just as they do when coexpressed in bacteria and that such an interaction activates the enzyme for cysteine desulfurase activity. The activation of Nfs1p was enhanced with increasing amounts of Isd11p added (Fig. 3B), and the persulfide formation was both PLP-(C) and temperature-dependent (D), just like in the case of the preformed and purified Nfs1p⅐Isd11p complex.
Isd11p-independent and Isd11p-dependent Steps in Persulfide Formation on Nfs1p-We sought to determine how Isd11p turns on Nfs1p activity. For brevity, the cysteine desulfurase activity of Nfs1p can be divided into two stages: binding of the substrate cysteine to the enzyme forming the cysteine-PLP ketimine adduct and subsequent formation of persulfide at the active site cysteine of the enzyme. Isd11p may be required for one or both of these steps. To address this issue, we developed two-step assays that allowed us to distinguish substrate cysteine binding from persulfide formation. In the first step, different amounts of Nfs1p were incubated with [ 35 S]cysteine in the absence of Isd11p. Nfs1p was then precipitated with ammonium sulfate to remove excess and unbound [ 35 S]cysteine. In the second step, Nfs1p with bound substrate was solubilized with buffer and incubated without or with different amounts of Isd11p. Samples were analyzed by non-reducing SDS-PAGE followed by autoradiography.  2 with lanes 3-6). Furthermore, the Nfs1p-associated 35 S signal was dependent on both Nfs1p (first step) and Isd11p (second step) amounts. These results suggest that Nfs1p can productively bind cysteine independent of Isd11p because [ 35 S]cysteine was included only in the first step and not in the second step and Isd11p was present only in the second step and not in the first step. Note that Nfs1p with [ 35 S]cysteine bound in the substrate-binding site probably cannot be detected by SDS gels because SDS releases the radioactive substrate. More importantly, although Isd11p was not required for Nfs1p to bind the substrate cysteine, it was essential for persulfide formation from bound cysteine. Isd11p may induce a conformational change in Nfs1p to bring the active site cysteine and the bound substrate cysteine closer, thereby facilitating the persulfide formation. Additional experiments were performed to further validate these conclusions, as described below.
For the two-step experiment in Fig. 4A, PLP was included in both the steps. Therefore, one set of control experiments involved PLP dependence of the two steps separately. Briefly, Nfs1p was incubated with [ 35 S]cysteine in the absence or presence of PLP (first step). Following removal of excess [ 35 S]cysteine and PLP, Isd11p was added and further incubated with or without added PLP (second step) (Fig. 4B). Efficient persulfide formation was detected only when PLP was included in the first step, regardless of PLP addition during the second step. These results reconfirm that Nfs1p by itself can bind cysteine in the absence of Isd11p but that PLP must be present. The substrate cysteine thus bound to the enzyme can be utilized for persulfide generation in the presence of Isd11p, and this process can occur with no further addition of PLP. In a different experiment (Fig.  4C), addition of Nfs1p by itself was found to inhibit [ 35 S]persulfide formation by the Nfs1p⅐Isd11p complex in a dose-dependent manner. Most likely, Nfs1p alone efficiently bound [ 35 S]cysteine, thereby making the substrate unavailable for the complex to form persulfide. These results further substantiate our conclusion that Nfs1p by itself can bind cysteine but cannot form persulfide and that the enzyme must be activated by Isd11p for persulfide generation.
Isd11p Protects the Active Site Cysteine of Nfs1p-Isd11p does not contain any cysteine residues. On the other hand, cysteine desulfurases form persulfide at the active site cysteine residue, and modification of the active site cysteine by an alkylating agent inactivates the enzyme (4). This observation prompted us to determine whether the presence of Isd11p has any effect on the active site cysteine of Nfs1p. For this purpose, we first optimized our assay conditions using the well studied A. vinelandii NifS as the model enzyme. The purified bacterial enzyme is highly active, demonstrating about three to four times more activity than the Nfs1p⅐Isd11p complex as assessed by [ 35 S]persulfide formation (no reducing agent present) or by spectrophotometric assay for sulfide release (DTT present). NifS was found previously to be highly sensitive to NEM treatment using the spectrophotometric assay (4). For our assays, bacterially expressed and purified A. vinelandii NifS was incubated with increasing concentrations of NEM, and then excess NEM was neutralized with DTT (first step). The enzyme was precipitated with ammonium sulfate to remove NEM and DTT. The protein pellet was solubilized with buffer, incubated with [ 35 S]cysteine and PLP (second step), and analyzed for [ 35 S]persulfide formation. In agreement with the previous report (4), NifS was found to be strongly inhibited by NEM at a concentration as low as 0.05 mM and was almost completely inhibited at 0.1 mM (Fig.  5A).
We then performed a similar two-step experiment with the Nfs1p⅐Isd11p complex (Fig. 5B, lanes 1-6) or Nfs1p by itself (lanes 7-12). During the second step of the assay, reaction mixtures containing NEM -treated or NEM-untreated Nfs1p alone were supplemented with Isd11p but not the corresponding Nfs1p⅐Isd11p samples. The results were striking. Nfs1p by itself was much more sensitive to NEM treatment than the Nfs1p⅐Isd11p complex. For example, Nfs1p treated with 1 mM NEM followed by Isd11p addition failed to generate any detectable persulfide (Fig. 5B, lane 10). In contrast, the preformed Nfs1p⅐Isd11p complex was only slightly affected by 1 mM NEM treatment (Fig. 5B, lane 4). These results suggest that Isd11p activates Nfs1p in a way that protects the active site cysteine from being modified by NEM. A conformational change in the enzyme would be in agreement with this notion.
An Isd11p Mutant Binds to Nfs1p but Fails to Activate the Enzymatic Activity and to Protect the Active Site Cysteine-Isd11p belongs to the LYR family of proteins that contain the conserved tripeptide LYR/K (18,19). Other members of this family include the B12 and B44 subunits of mitochondrial complex I of the respiratory chain of various organisms. However, the role of the tripeptide motif in any of these proteins is unknown. We mutated the tripeptide 15 LYK 17 of Isd11p to 15 AAA 17 , and the phenotypic defects that occur from this mutation in S. cerevisiae will be published elsewhere. 3 The corresponding protein (Isd11p mut) and Nfs1p-His 6 were coexpressed in bacteria in soluble forms. Like in the case of authentic Isd11p, the Isd11p mut protein also formed a complex (Nfs1p⅐Isd11p mut) as it was copurified with Nfs1p-His 6 by Ni-NTA chromatography. The affinities of Isd11p and the LYK mutant proteins for binding to Nfs1p remain to be determined. Interestingly, the composition of protein components in the Nfs1p⅐Isd11p and Nfs1p⅐Isd11p mut complexes was very similar (Fig. 6A). Unlike in the case for Nfs1p⅐Isd11p, however, the Nfs1p⅐Isd11p mut complex failed to exhibit any cysteine desulfurase activity (Fig.  6B, lane 1). Most likely, the Isd11p mutant interacts with Nfs1p but cannot induce the conformational change in the enzyme necessary for persulfide formation. Interestingly, addition of Isd11p to the Nfs1p⅐Isd11p mut complex restored the activity in a dose-dependent manner (Fig. 6B, lanes 2-4). Higher concentrations of Isd11p may replace the Isd11p mutant protein in the complex, thereby leading to the conformational change in Nfs1p required for its activity.
To further validate these notions, we compared NEM sensitivity of the two complexes: Nfs1p⅐Isd11p versus Nfs1p⅐Isd11p mut (Fig. 6C). The latter was much more sensitive than the former. For example, pretreatment with 1 mM NEM mostly inhibited the mutant complex but only slightly the authentic complex. With regard to NEM (1 mM) sensitivity, the Nfs1p⅐Isd11p mut practically behaved like Nfs1p alone. These results strongly correlate the role of Isd11p in enzyme activation with active site cysteine protection. Mere binding of Isd11p is not sufficient for either of these two phenomena. We conclude that interaction of Isd11p with Nfs1p induces a conformational change in the enzyme that is essential for the cysteine desulfurase activity and that the active site cysteine residue is not easily accessible to alkylating agents when the enzyme is activated.

DISCUSSION
Nfs1p is the yeast homolog of the bacterial cysteine desulfurases NifS and IscS, with a conserved PLP-binding domain, active site cysteine residue, and other features (36). Nfs1p is primarily localized in mitochondria (11) and is known to interact with a small protein, Isd11p (18,19). Isd11p is found in eukaryotes but not in prokaryotes. Here we describe a physiological function for this eukaryote-specific Isd11p interaction with Nfs1p. Surprisingly, we found that, unlike bacterial cysteine desulfurases, the mitochondrial enzyme Nfs1p is completely inactive on its own and requires Isd11p for its activity. For the first time we present data on the active mitochondrial enzyme complex, showing a requirement for both Nfs1p and Isd11p. We have also identified Isd11p-independent and Isd11p-dependent steps involved in the Nfs1p-mediated cysteine desulfuration reaction.
The key results of this study and a model based on these data are schematically presented in Fig. 7. First, bacterially expressed and purified Nfs1p⅐Isd11p complex exhibited strong cysteine desulfurase activity. By contrast, Nfs1p by itself was completely inactive (Fig. 2, A and F). Second, addition of Isd11p activated the Nfs1p cysteine desulfurase activity in a dose-dependent manner (Fig. 3B). Furthermore, this activity was similar to that of the preformed Nfs1p⅐Isd11p complex with respect to protein 3 A. Pandey, R. Golla, A. Dancis, and D. Pain, manuscript in preparation. amounts (Fig. 3A) and also PLP and temperature dependence (C and D). Third, Nfs1p by itself was able to bind the substrate cysteine in a PLP-dependent manner but failed to form the persulfide (Fig. 4, A and B). In fact, Nfs1p efficiently competed with the Nfs1p⅐Isd11p complex for binding to cysteine (Fig. 4C). Remarkably, Nfs1p with prebound cysteine was able to efficiently form the persulfide with the addition of Isd11p (Fig. 4A). On the basis of these results, we conclude that Isd11p is not required for Nfs1p binding to cysteine (Isd11p-independent step). However, Isd11p is essential for the subsequent step, i.e. formation of the persulfide at the active site cysteine from the bound substrate cysteine (Isd11p-dependent step).
We propose that in the absence of Isd11p, the active site cysteine residue 385 of Nfs1p is not close enough to act on the substrate cysteine bound to PLP at lysine residue 263 and thus fails to form the persulfide. A critical interaction of Isd11p with Nfs1p induces a conformational change in the enzyme that brings the active site cysteine and the substrate cysteine closer, thereby facilitating persulfide formation (Fig. 7). Primary sequence alignment reveals that the mature form of yeast Nfs1p contains a stretch of N-terminal amino acids that is absent in A. vinelandii NifS. These extra N-terminal amino acids in Nfs1p could be critical for enzyme activation perhaps through interactions with Isd11p. The notion of Isd11p-mediated conformational change in the enzyme is supported by other independent observations. For example, Isd11p with a mutation in the conserved LYK motif interacted with Nfs1p but failed to activate the enzyme (Fig. 6, A and B). This could be due to the inability of the Isd11p mutant protein to induce the necessary conformational change in the enzyme. Mere binding of the Isd11p mutant was not sufficient to activate the enzyme.
Another supporting evidence for a conformational change in the enzyme is suggested by the differential NEM sensitivity of the active site cysteine. Compared with Nfs1p alone, the Nfs1p⅐Isd11p complex was more resistant to inactivation by NEM (Fig. 5B). Isd11p may mediate the conformational change in Nfs1p in such a way that the active site cysteine becomes less accessible to modification by NEM. The Isd11p mutant was less effective in protecting the active site cysteine from NEM modification (Fig. 6C), possibly because of its inability to induce the conformational change in the enzyme. From what is known about the bacterial IscS structure (9), the movement of the peptide loop containing the active site cysteine toward the substrate-binding site might very well generate a more compact structure that would better shield the active site from NEM. Another step in the enzymatic cycle of cysteine desulfurases involves sulfur donation to recipients such as the Isu Fe-S cluster scaffold or other proteins involved in tRNA thiolation. The active site loop in the enzyme might need to undergo a second conformational change, in this case moving away from the substrate-binding site and toward the sulfur acceptor. The role of Isd11p in this secondary change has not been examined. Much work will be needed to directly demonstrate the conformational changes involved in Nfs1p activity. For this purpose, structures of the Nfs1p⅐Isd11p and Nfs1p⅐Isd11p mutant complexes would be extremely important. A comparison of these structures with those of bacterial cysteine desulfurases will be equally important to determine fundamental mechanistic differences in eukaryotic and prokaryotic enzymes. In any case, the data presented here clearly show that Isd11p activates the Nfs1p cysteine desulfurase activity.
Previous studies (18,19) have overlooked the requirement of Isd11p for Nfs1p activity. Mitochondria isolated from a temperature-sensitive isd11 mutant or Isd11p-depleted cells exhibited Fe-S cluster deficiency. When these mutant mitochondrial extracts were incubated with cysteine and DTT, sulfide produc-tion was found to be unaffected or even slightly enhanced. On the basis of these results, it was concluded that Isd11p is not required for the cysteine desulfurase activity of Nfs1p, rather it functions as a stabilizer of Nfs1p. However, no evidence was presented demonstrating that the observed sulfide production in these mutant mitochondria was indeed due to Nfs1p and not due to activities of other proteins/enzymes. In fact, using the same assay, we have seen sulfide production not only in Isd11pdepleted mitochondria but also in Nfs1p-depleted mitochondria. The activity, being present in Nfs1p-depleted mitochondria, cannot be attributed to the Nfs1p cysteine desulfurase. In Isd11p-depleted mitochondria, Nfs1p fails to form [ 35 S]persulfide at its active site. These results are in agreement with the data presented here and will be published elsewhere.
Fe-S cluster biogenesis in mitochondria is an essential and conserved process. It is highly complex, and the mechanistic details remain to be determined. The steps involved must be tightly regulated because both components (iron and sulfur) are toxic when unassembled or present in excess. The Nfs1p⅐Isd11p complex interacts with the scaffold proteins FIGURE 7. Model for Isd11p-mediated activation of Nfs1p for persulfide formation. A schematic is presented depicting how Isd11p might be involved in Nfs1p activation. The PLP-conjugating lysine 263 and the active site cysteine 385 of Nfs1p correspond to the mature form of the protein in mitochondria (11,32). See text for details.
(Isu1p/Isu2p) for the synthesis of Fe-S cluster intermediates, with the sulfur provided by Nfs1p and the iron donated by the yeast frataxin homolog Yfh1p (23,25). Recent studies suggest that frataxin interacts with a complex of human Nfs1, Isd11, and Isu2 and serves as an allosteric switch regulating the synthesis of Fe-S clusters (37,38). The data presented here may point to another level of regulation. For example, Isd11p in mitochondria may reversibly interact with Nfs1p, depending on the demand for Fe-S cluster synthesis, thereby turning the cysteine desulfurase activity on and off. This could be mediated by the availability of mitochondrial cysteine, iron, and/or nucleotides (ATP, GTP, or NADPH) required for Fe-S cluster biogenesis (28,29,39). The distribution of Isd11p between the mitochondrial inner membrane and the soluble matrix (18) might also be important for localizing the active cysteine desulfurase to critical locations within mitochondria. Proper regulation of the cysteine desulfurase activity and the role of Isd11p are likely to be relevant to vital cellular functions, including the tricarboxylic acid cycle, respiration, and protein translation.