Cysteine desulfurases abstract sulfur from the substrate cysteine, generate a covalent persulfide on the active site cysteine of the enzyme, and then donate the persulfide sulfur to various recipients such as Fe–S clusters. In Saccharomyces cerevisiae, the Nfs1p protein is the only known cysteine desulfurase, and it forms a complex with Isd11p (Nfs1p·Isd11p). Both of these proteins are found primarily in mitochondria and both are essential for cell viability. In the present study we show, using the results of experiments with isolated mitochondria and purified proteins, that Isd11p is required for the cysteine desulfurase activity of Nfs1p. Whereas Nfs1p by itself was inactive, the Nfs1p·Isd11p complex formed persulfide and was active as a cysteine desulfurase. In the absence of Isd11p, Nfs1p was able to bind the substrate cysteine but failed to form a persulfide. Addition of Isd11p allowed Nfs1p with bound substrate to generate a covalent persulfide. We suggest that Isd11p induces an activating conformational change in Nfs1p to bring the bound substrate and the active site cysteine in proximity for persulfide formation. Thus mitochondrial Nfs1p is different from bacterial cysteine desulfurases that are active in the absence of accessory proteins. Isd11p may serve to regulate cysteine desulfurase activity in mitochondria.

INTRODUCTION

Fe–S clusters are modular cofactors consisting of iron and sulfur liganded to cysteine sulfurs of proteins. They function in essential processes, such as catalysis, electron transfer, structural stabilization of protein domains and in sensing of environmental signals (oxygen, superoxide, iron) [13]. Fe–S cluster biogenesis is a complex process involving more than a dozen genes/proteins [4,5]. Lethal phenotypes are associated with disruption of many of these genes, consistent with the essential nature of the process. In prokaryotes, three distinct operons – nif, isc and suf – encode proteins that participate in Fe–S cluster synthesis. The Nif (nitrogen fixation) system is specialized for high-flux Fe–S cluster assembly on nitrogenase in nitrogen-fixing organisms such as Azotobacter vinelandii. The Isc (Fe–S cluster) system constitutes the ‘housekeeping’ Fe–S cluster machinery for bacteria including Escherichia coli and A. vinelandii. The Suf (sulfur mobilization) system is specially adapted for conditions such as oxidative stress and iron deprivation that render the Isc machinery inadequate. The initial step in each of these three systems involves mobilization of sulfur from cysteine by specific cysteine desulfurases NifS, IscS or SufS respectively [68]. A persulfide intermediate formed on the cysteine desulfurase as part of the enzymatic reaction is then transferred to a scaffold protein for incorporation into an Fe–S cluster intermediate. Subsequently, the Fe–S cluster intermediate is transferred to apoproteins, forming the holo or active proteins.

Cysteine desulfurases abstract sulfur from the amino acid cysteine and deliver it to recipients. The enzymatic reaction occurs in three major steps. The substrate cysteine binds to an enzyme pocket containing the cofactor PLP (pyridoxal phosphate). The active site cysteine residue of the enzyme initiates a nucleophilic attack on the sulfhydryl group of the substrate cysteine–PLP adduct, forming a covalent persulfide at the active site cysteine residue [8]. The persulfide is then transferred to designated proteins involved in Fe–S cluster synthesis or thio-modification of tRNAs. Structures of prokaryotic cysteine desulfurases have been solved, although, to date, no eukaryotic structures are known. NifS orthologues are homodimers, with each monomer containing two domains – the larger containing a PLP/substrate-binding site and the smaller carrying the active site cysteine residue [911]. In the structure of the E. coli IscS, the loop containing the active site cysteine residue is disordered and is found more than 17 Å (1 Å=0.1 nm) away from the PLP cofactor and the substrate-binding site. A large conformational change in this region may therefore be required to bring the loop close to the substrate during enzyme catalysis. Purified A. vinelandii NifS exhibits constitutively high cysteine desulfurase activity, and this activity does not require any other accessory proteins. The enzyme is extremely susceptible to alkylating agents such as NEM (N-ethylmaleimide) owing to inactivation of the active site cysteine residue [12,13].

In eukaryotes, orthologues of many of the components involved in Fe–S cluster assembly in prokaryotes are found in mitochondria, consistent with the evolutionary origin of the organelle [5,14]. In the yeast Saccharomyces cerevisiae, only one gene (NFS1) coding for a cysteine desulfurase protein exists, and the gene product (Nfs1p) is found primarily in mitochondria [15]. A small but functionally essential portion of extra-mitochondrial Nfs1p also exists, which is perhaps involved in tRNA thiolation and synthesis or repair of some Fe–S clusters outside mitochondria [1618]. The Nfs1p cysteine desulfurase is required for Fe–S cluster synthesis in mitochondria and thus for activities of Fe–S cluster proteins in the organelle. If isolated and intact mitochondria are supplied with [35S]cysteine, they are able to efficiently synthesize and assemble Fe–35S clusters on an endogenous apoprotein (e.g. apoaconitase) or on a newly imported apoprotein (e.g. apoferredoxin) [1922]. This biosynthetic activity requires the cysteine desulfurase Nfs1p in mitochondria [19]. Nfs1p associates with a small mitochondrial protein, Isd11p (11 kDa), forming a complex (Nfs1p·Isd11p) [23,24]. Similar to Nfs1p, Isd11p is also essential for cell viability. Whereas cysteine desulfurases are highly conserved, Isd11p is found only in eukaryotes and not in prokaryotes, suggesting a unique and novel function of the protein in mitochondria. Cells with a temperature-sensitive isd11 allele or cells lacking Isd11p are deficient in Fe–S cluster protein activities [23,24], implicating Isd11p with Nfs1p and other proteins in the process of Fe–S cluster assembly. Similar to the case of the yeast Isd11p, knockdown of the corresponding homologue in HeLa cells also results in the deficiency of Fe–S cluster proteins [25].

However, a specific role of Isd11p in the enzymatic function of Nfs1p, i.e. in mediating or regulating the cysteine desulfurase activity, has not yet been ascertained. Two types of assays were previously utilized. Extracts of mitochondria isolated from isd11 mutant cells were supplemented with cysteine and DTT (dithiothreitol), and sulfide production was measured. In this assay, there was no effect of Isd11p depletion or inactivation on mitochondrial sulfide production [23,24]. These results led to the conclusion that Isd11p is not required for the cysteine desulfurase activity of Nfs1p, and that its principal function may be to stabilize Nfs1p protein. However, there was no evidence that the observed sulfide production in these mutant mitochondria was really due to Nfs1p activity. The possibility of Nfs1p-independent sulfide production by other proteins/enzymes or processes in mitochondria was not tested. The role of Isd11p in mitochondrial cysteine desulfurase activity therefore remained inconclusive. A different assay involved incubation of a detergent lysate of mitochondria with [35S]cysteine followed by Blue Native gel separation of the Nfs1p·Isd11p complex. The complex was found to be radiolabelled by 35S, suggesting activity of the complex towards cysteine [23]. However, Nfs1p by itself was not detected in the Blue Native gel, and so a specific role for Isd11p in Nfs1p activity in mitochondria could not be ascertained. In summary, the function of Isd11p in mitochondria has not been determined, and the link between Isd11p depletion/inactivation and Fe–S cluster deficiency in mitochondria has remained elusive.

To study the Nfs1p cysteine desulfurase activity in mitochondria, we have recently developed a different type of radioactive assay [26]. When isolated mitochondria were supplemented with [35S]cysteine, a radiolabelled persulfide associated with Nfs1p was identified. The 35S label detected on Nfs1p reflected the cysteine desulfurase activity of the enzyme and acted as a bona fide biochemical intermediate, varying reciprocally with conditions that block or allow Fe–S cluster synthesis in mitochondria. In the present study we show that formation of a covalent persulfide on Nfs1p and Fe–S cluster synthesis in mitochondria were entirely dependent on the presence of the Isd11p accessory protein. In the absence of Isd11p, Nfs1p bound to the substrate [35S]cysteine but did not form the persulfide. The interaction of Isd11p with Nfs1p was required for covalent persulfide (Nfs1p-S–35SH) formation, perhaps by inducing a conformational change that brought the active site cysteine and the bound substrate into proximity. These results obtained with isolated mitochondria are in good agreement with experiments using purified proteins, showing that Nfs1p by itself was inactive and required Isd11p for its activity. Thus the mitochondrial cysteine desulfurase, by requiring an accessory protein for its activity, is different from the prokaryotic IscS or NifS, which do not require any interacting proteins for activity.

MATERIALS AND METHODS

Yeast strains and growth conditions

Regulated expression of Nfs1p from the GAL10 promoter in the YPH499 background was achieved as described previously [15,19]. Regulated expression of Isd11p was accomplished by use of a promoter swap introduced into YPH499 using a PCR product amplified from the pFA6a-His3MX6-PGAL1 template and targeted to the ISD11 locus as described previously [27]. Correct integration at ISD11 and regulated expression of Isd11p were confirmed. The Gal-Isd11p strain was grown at 30°C in the presence of galactose (0.5%) in raffinose-based medium (6.7 g/l yeast nitrogen base without amino acids, 2 g/l complete medium supplements minus uracil, and 2% raffinose). For depletion of Isd11p, cells were shifted to the same raffinose-based medium without galactose and grown at 30°C for different time periods. Cultures of the WT (wild-type) parental strain YPH499 were handled in a similar fashion.

The genomic region of ISD11, including 700 bp of the 5′ flanking region, the ORF (open reading frame) and 200 bp of the 3′ flanking region was amplified by PCR with EagI (5′) and SalI (3′) ends. The fragment was inserted into the corresponding restriction sites of pRS416, generating the plasmid pRS416/Isd11p. The amino acids L15YK17 were then changed to A15AA17 using the QuikChange® site-directed mutagenesis kit (Stratagene), generating the plasmid pRS416/Isd11p mut. Plasmids pRS416/Isd11p, pRS416 (empty) and pRS416/Isd11p mut were introduced into the Gal-Isd11p strain. The transformants were grown in defined medium minus uracil, and the carbon source was shifted from galactose to raffinose or glucose, down-regulating expression of the genomic ISD11 and uncovering the phenotypes associated with the plasmids. Mitochondria from various strains were isolated as described previously [20].

Bacterial expression of proteins

The mature (signal-cleaved) form of Nfs1p in mitochondria lacks the N-terminal 36 amino acids of the corresponding nuclear-encoded precursor protein [18]. Using appropriate primers, the ORF for mature Nfs1p was amplified by PCR from a yeast genomic library to obtain the NdeI-ORF-XhoI product. Unlike Nfs1p, Isd11p does not contain a cleavable mitochondrial targeting signal [23,24], and the entire ISD11 ORF was amplified in a similar manner. After digestion with NdeI and XhoI, the PCR products were cloned into the same sites of pET21b (Novagen), thereby introducing a His6 tag in-frame at the C-terminus of the proteins. To obtain the Nfs1p·Isd11p complex formed in vivo in bacteria, we generated a construct that contained a T7 promoter driving a polycistronic mRNA for mature Nfs1p with a C-terminal His6 tag (Nfs1p–His6) and Isd11p, each with separate ribosome-binding sites [26]. The plasmid pT7-7/NifS–His6 was generated for expression of A. vinelandii NifS as described previously [26]. NifS–His6, Nfs1p·Isd11p and Isd11p–His6 were expressed in BL21 (DE3) Codon Plus cells (Stratagene), and A. vinelandii NifS–His6 was expressed in BL21 (DE3) cells.

Expression of Nfs1p–His6 alone or co-expression of Nfs1p–His6 and Isd11p was induced with 0.5 mM IPTG (isopropyl β-D-thiogalactopyranoside) for 20 h at 25°C. A. vinelandii NifS–His6 was expressed with 0.5 mM IPTG for 3 h at 37°C. These proteins were purified essentially as described previously [26]. Briefly, cells were treated with lysozyme (50 μg/ml) in buffer A [50 mM Tris/HCl, pH 8.0, 0.15 M NaCl, 10% (v/v) glycerol and 1 mM PMSF] for 30 min on ice, and then disrupted using a probe sonicator (Branson Sonifier 450). Cell lysates were centrifuged at 12000 g for 30 min at 4°C, and the supernatant fraction was incubated with Ni-NTA (Ni2+-nitrilotriacetate) agarose for 3 h at 4°C. The resin was washed with buffer A containing 10 mM imidazole. Bound proteins were then eluted with 0.4 M imidazole in buffer A and stored at −80°C. The yields of purified Nfs1p–His6 and Nfs1p·Isd11p proteins were 90 μg and 100 μg respectively from 100 ml of starting cultures. For A. vinelandii NifS–His6, the yield was 2.5 mg per 100 ml of starting culture.

The protocol for Isd11p–His6 expression and purification was modified from that of the other proteins because of the poor solubility of the overexpressed protein. Specifically, BL21 (DE3) Codon Plus cells containing the plasmid pET21b/Isd11p-His6 were grown at 37°C in LB (Luria–Bertani) broth containing 100 μg/ml ampicillin and 34 μg/ml chloramphenicol to a D600 of ~0.8. The culture was shifted to 25°C and sorbitol was added to a final concentration of 0.5 M to enhance solubility of the expressed protein. Protein expression was initiated with 0.2 mM IPTG and continued for 72 h at 25°C. Cells were harvested and washed with buffer B [50 mM Tris/HCl, pH 8.0, 10% (v/v) glycerol and 1 mM PMSF]. After resuspension in the same buffer, cells were disrupted with a probe sonicator as indicated above. The cell lysate was centrifuged at 12000 g for 30 min at 4°C. Under these conditions, approximately 10% of the total expressed protein was found in the supernatant fraction. The supernatant fraction containing soluble Isd11p–His6 was incubated with Ni-NTA agarose for 3 h at 4°C. The resin was washed with buffer B, and bound proteins were eluted with 0.4 M imidazole in buffer B and stored at −80°C. The yield of purified and soluble Isd11p–His6 was 8–10 μg from 100 ml of bacterial culture. The non-denatured (or native) Isd11p thus purified was used for all experiments involving radiolabelled persulfide formation using mitochondrial lysates or purified Nfs1p protein, and not for import into mitochondria.

For assays involving mitochondrial import, urea-denatured Isd11p was used, as it was likely to be unfolded and therefore rapidly imported [21,28]. Briefly, BL21 (DE3) Codon Plus cells, carrying the plasmid pET21b/Isd11p-His6, were cultured at 37°C in LB broth supplemented with 100 μg/ml ampicillin and 34 μg/ml chloramphenicol to a D600 of ~0.8. Following addition of 1 mM IPTG, protein expression was continued for 3 h at 37°C. Under these conditions, Isd11p protein was found exclusively in the inclusion bodies. The overexpressed protein was solubilized with 8 M urea in 50 mM Tris/HCl, pH 8.0, and centrifuged at 250000 g for 20 min at 20°C to remove insoluble materials as described for other mitochondrial proteins expressed in bacteria [19,28,29]. The supernatant fraction contained Isd11p with more than 80% purity, and the protein was not further purified by Ni-NTA agarose chromatography. The yield was approximately 500 μg of Isd11p protein from 100 ml of starting culture. The high yield of Isd11p from inclusion bodies allowed us to use an adequate amount (5 μg) of the protein per reaction in sulfide-forming cysteine desulfurase assays (Figure 2D). For some experiments (Supplementary Figure S3), radiolabelled Isd11p was desired and hence protein induction was carried out in M9 media containing 100 μg/ml ampicillin, 34 μg/ml chloramphenicol, 25 μCi/ml [35S]methionine (1175 Ci/mmol) and 1 mM IPTG at 37°C for 3 h. Inclusion bodies containing radiolabelled Isd11p as the major protein were processed as described above for the unlabelled protein. Both unlabelled and radiolabelled Isd11p were stored in 50 mM Tris/HCl, pH 8.0, containing 8 M urea at −80°C until further use.

Cysteine desulfurase assay for sulfide production

The cysteine desulfurase activities of mitochondrial extracts or purified proteins were measured by a spectrophotometric assay as follows. Mitochondria were lysed with 0.2% dodecylmaltoside in 50 mM Tris/HCl, pH 8.0 [24]. Various amounts of mitochondrial lysate or purified proteins were added to sealed microfuge tubes containing 50 mM Tris/HCl, pH 8.0, 5 mM L-cysteine, 5 mM DTT and 0.2 mM PLP, and the final volume was 350 μl. After incubation at 37°C for 1 h, the reaction was stopped by the addition of 0.5% NaOH. Sulfide release was measured as described previously [30]. Briefly, 125 μl of 0.1% DPD (N,N-dimethyl-p-phenylenediamine dihydrochloride in 5 M HCl) and 50 μl of 11.5 mM FeCl3 were added, and incubated at 37°C for 20 min. Samples were centrifuged at 15000 g for 5 min at 25°C, and the absorbance of the supernatant was determined at 670 nm.

Formation of covalent [35S]persulfide on Nfs1p

Nfs1p in mitochondria

To deplete endogenous nucleotides, mitochondria in HS buffer (20 mM Hepes/KOH, pH 7.5, and 0.6 M sorbitol) were incubated at 30°C for 10 min [1921]. Typically, mitochondria (200 μg of protein) were mixed with 10 μCi of [35S]cysteine (1075 Ci/mmol) in HS buffer containing 10 mM magnesium acetate and 40 mM potassium acetate in a final volume of 50 μl. After incubation at 30°C for 15 min, reaction mixtures were diluted with HS buffer containing 0.15 M NaCl and kept on ice for 10 min. Mitochondrial pellets were isolated by centrifugation at 15000 g for 10 min at 4°C. Samples were analysed by non-reducing SDS/PAGE followed by autoradiography using Kodak Biomax MR films. Variations of the assay are described in the corresponding Figure legends.

Purified Nfs1p

Various amounts of purified proteins were added to HS buffer containing 0.15 mM PLP, 0.15 M NaCl and 10 μCi of [35S]cysteine. The final volume of the assay was 50 μl, and samples were incubated at 30°C for 15 min [26]. An equal volume of ice-cold 20% TCA (trichloroacetic acid) was added and samples were left on ice for 1 h. After centrifugation at 15000 g for 30 min at 4°C, the protein pellets were analysed by non-reducing SDS/PAGE followed by autoradiography. In some cases, proteins were treated with NEM and/or DTT prior to incubation with [35S]cysteine as indicated in the corresponding Figure legends.

Fe–S cluster synthesis assays

These assays were performed essentially as described previously [1922,26]. Briefly, isolated and intact mitochondria were incubated with [35S]cysteine in the presence of added nucleotides (4 mM ATP, 1 mM GTP and 2 mM NADH) and iron as ferrous ascorbate (10 μM). Insertion of newly formed Fe–35S clusters into endogenous apoaconitase was visualized by native PAGE, followed by autoradiography.

Miscellaneous procedures

Spectrophotometric assays for aconitase [31], succinate dehydrogenase [32] and malate dehydrogenase [33] activities have been described previously. Cellular iron uptake and mitochondrial iron levels were determined by growing cells in the presence of standard defined medium supplemented with radioactive iron (100 nM 55Fe ferrous ascorbate, 100 mCi/mg, CNL Scientific), followed by mitochondria isolation and scintillation counting [15]. Quantification of radiolabelled protein bands was performed by densitometric analysis of autoradiographs using NIH ImageJ software.

RESULTS

Assays for cysteine desulfurase activity in mitochondria

Previous studies by others used a spectrophotometric assay to determine the cysteine desulfurase activity in mitochondria [23,24]. In this assay, the mitochondrial extract was incubated with the amino acid cysteine (unlabelled) under reducing conditions, followed by measurement of multiple rounds of persulfide formation and release of persulfide sulfur as sulfide. To standardize this assay in our hands, we started with bacterially expressed and purified A. vinelandii NifS (Supplementary Figure S1A at http://www.BiochemJ.org/bj/448/bj4480171add.htm) as a positive control. The enzymatic properties of NifS have been extensively studied and the activity of purified NifS has been successfully determined by similar assays for sulfide production [12,13]. As expected, we also found purified NifS to be highly active in our spectrophotometric assays (Supplementary Figure S1B, bar 1). As an additional positive control, yeast Nfs1p–His6 was co-expressed with Isd11p in bacteria and the Nfs1p·Isd11p complex that formed in vivo was purified to more than 90% homogeneity (Supplementary Figure S1A) [26]. With respect to the molecular nature and mass (~200 kDa), the purified complex behaved very similarly to the complex in mitochondria [23,24,26]. As with the A. vinelandii NifS, the purified Nfs1p·Isd11p complex was also found to be active by the colorimetric assay (Supplementary Figure S1B, bar 2), although NifS exhibited 3–4-fold more activity.

Nfs1p is the only known cysteine desulfurase in mitochondria [15]. To be able to determine Nfs1p-specific sulfide production using the colorimetric assay, we generated a yeast strain that expresses Nfs1p from the regulated GAL10 promoter. Protein expression was turned off by shifting the cells from inducing (galactose-raffinose) to non-inducing (raffinose) medium [15,26]. Cells were grown in raffinose-based medium for 22 h at 30°C and mitochondria were isolated. These mitochondria [Nfs1p▼ (Nfs1p-depleted)] did not contain any detectable Nfs1p as judged by immunoblotting. Isd11p protein levels remained practically unaffected under these conditions, thereby serving as an internal loading control (Supplementary Figure S1C) [26]. Furthermore, Nfs1p▼ mitochondria exhibited iron-related phenotypes, including iron accumulation, deficiency of Fe–S cluster synthesis and loss of Fe–S cluster-containing enzyme activities, consistent with the fundamental role of Nfs1p in Fe–S cluster biogenesis [15,19]. However, experiments to determine the sulfide-forming activity of Nfs1p▼ mitochondria yielded completely unexpected results. Cysteine desulfurase activity as measured by the colorimetric assay was found to be unaffected and even slightly increased in the mitochondrial lysate depleted of Nfs1p (Supplementary Figure S1B, compare bars 3 and 4 with bars 5 and 6 respectively). The sulfide-generating activity in mitochondria with no detectable Nfs1p cannot be attributed to the Nfs1p cysteine desulfurase. These results suggest the existence of background mitochondrial activities capable of releasing sulfur from cysteine, which are independent of Nfs1p and lack functional redundancy with Nfs1p. In mitochondria, Nfs1p is not an abundant protein and it appears to be present at less than 0.01% of total proteins [26]. The colorimetric assay required a minimum of 2–5 μg of purified enzymes; it is too insensitive and could not be reliably used for our studies with cysteine desulfurase activity in mitochondria. A different assay was therefore needed, and we have recently optimized a radioactive assay able to detect covalent [35S]persulfide formation on cysteine desulfurases [26]. Unlike in the case of the spectrophotometric assay for continuous sulfide release (DTT present), the radioactive assay detects persulfide formation from a single round of activity (DTT absent). The radioactive assay is at least 50–100-fold more sensitive than the spectrophotometric assay and hence it was used for most of the experiments of the present study as described below.

Requirement of Isd11p for Nfs1p-bound persulfide formation in mitochondria

We wanted to test the function of Isd11p in Nfs1p-dependent cysteine desulfurase activity in mitochondria. In order to do so, we sought to identify mitochondria containing Nfs1p but lacking Isd11p. The Gal-Isd11p strain was shifted from an inducing carbon source (galactose) to a non-inducing carbon source (raffinose), and Nfs1p and Isd11p protein levels were monitored. Isd11p was rapidly depleted, consistent with a short half-life (less than 3 h) of the protein (Figure 1A, bottom panel). Nfs1p protein levels were also decreased (Figure 1A, top panel) and became undetectable after 16 h (results not shown). However, a time window could be identified during which Nfs1p was present and Isd11p was severely depleted (Figure 1A, top and bottom panels, lanes 7 and 8). Mitochondria were isolated from WT and Gal-Isd11p cells 9 h after a shift to raffinose [Isd11p▼ (Isd11p-depleted)].

Nfs1p cannot form persulfide in Isd11p▼ mitochondria and Isd11p imported into these mitochondria restores the process

Figure 1
Nfs1p cannot form persulfide in Isd11p▼ mitochondria and Isd11p imported into these mitochondria restores the process

(A) The Gal-Isd11p strain was grown at 30°C in raffinose-based medium containing galactose. For depletion of Isd11p, cells were shifted to the same raffinose-based medium without galactose and grown at 30°C for the indicated time periods (Isd11p▼). WT control cells were treated in the same way. Mitochondrial extracts were analysed by SDS/PAGE under reducing conditions followed by immunoblotting using anti-Nfs1p (top panel) and anti-Isd11p (bottom panel) antibodies. (B) Mitochondria were isolated from the Gal-Isd11p strain after 9 h of depletion of Isd11p in galactose-free medium as in (A) above. These Isd11p▼ mitochondria in HS buffer were pre-incubated with or without valinomycin (5 μM) for 5 min on ice. Samples were then adjusted so that mitochondria (200 μg of protein) were in HS buffer containing 10 mM magnesium acetate, 40 mM potassium acetate, 4 mM ATP, 1 mM GTP, 2 mM NADH and 0.1 mg/ml BSA. Reaction mixtures were incubated at 30°C for 15 min, with or without addition of 100 ng of urea-denatured Isd11p (1st step). The total volume was 50 μl, and the final urea concentration was 160 mM. Samples were diluted 20-fold with HS buffer, and centrifuged at 15000 g for 5 min at 4°C. The mitochondrial pellets (with or without imported Isd11p) thus obtained were resuspended in HS buffer containing 10 mM magnesium acetate and 40 mM potassium acetate, incubated with 10 μCi of [35S]cysteine for 15 min at 30°C, and assayed for persulfide formation (2nd step) in the absence or presence of added nucleotides (4 mM ATP, 1 mM GTP and 2 mM NADH) and ferrous ascorbate (Fe2+; 10 μM). Reaction mixtures were diluted with HS buffer containing 0.15 M NaCl, kept on ice for 10 min, and mitochondrial pellets were re-isolated by centrifugation. WT and Nfs1p▼ mitochondria served as positive and negative controls respectively. As another positive control, the purified Nfs1p·Isd11p complex (50 ng) was incubated with PLP (0.15 mM) and [35S]cysteine (10 μCi) in HS buffer containing 0.15 M NaCl in a final volume of 50 μl. After incubation at 30°C for 10 min, proteins were precipitated with 10% TCA. All samples were analysed by non-reducing SDS/PAGE followed by autoradiography. The intensity of Nfs1p-S–35SH in WT mitochondria in the absence of added nucleotides and iron (lane 1) was considered 100%. Arrowheads indicate bands (24 and 22 kDa) present in lanes 1, 3, 4 and 6. The asterisk indicates a major band (12 kDa) present in lanes 1 and 3. Molecular masses in kDa are indicated to the left-hand side.

Figure 1
Nfs1p cannot form persulfide in Isd11p▼ mitochondria and Isd11p imported into these mitochondria restores the process

(A) The Gal-Isd11p strain was grown at 30°C in raffinose-based medium containing galactose. For depletion of Isd11p, cells were shifted to the same raffinose-based medium without galactose and grown at 30°C for the indicated time periods (Isd11p▼). WT control cells were treated in the same way. Mitochondrial extracts were analysed by SDS/PAGE under reducing conditions followed by immunoblotting using anti-Nfs1p (top panel) and anti-Isd11p (bottom panel) antibodies. (B) Mitochondria were isolated from the Gal-Isd11p strain after 9 h of depletion of Isd11p in galactose-free medium as in (A) above. These Isd11p▼ mitochondria in HS buffer were pre-incubated with or without valinomycin (5 μM) for 5 min on ice. Samples were then adjusted so that mitochondria (200 μg of protein) were in HS buffer containing 10 mM magnesium acetate, 40 mM potassium acetate, 4 mM ATP, 1 mM GTP, 2 mM NADH and 0.1 mg/ml BSA. Reaction mixtures were incubated at 30°C for 15 min, with or without addition of 100 ng of urea-denatured Isd11p (1st step). The total volume was 50 μl, and the final urea concentration was 160 mM. Samples were diluted 20-fold with HS buffer, and centrifuged at 15000 g for 5 min at 4°C. The mitochondrial pellets (with or without imported Isd11p) thus obtained were resuspended in HS buffer containing 10 mM magnesium acetate and 40 mM potassium acetate, incubated with 10 μCi of [35S]cysteine for 15 min at 30°C, and assayed for persulfide formation (2nd step) in the absence or presence of added nucleotides (4 mM ATP, 1 mM GTP and 2 mM NADH) and ferrous ascorbate (Fe2+; 10 μM). Reaction mixtures were diluted with HS buffer containing 0.15 M NaCl, kept on ice for 10 min, and mitochondrial pellets were re-isolated by centrifugation. WT and Nfs1p▼ mitochondria served as positive and negative controls respectively. As another positive control, the purified Nfs1p·Isd11p complex (50 ng) was incubated with PLP (0.15 mM) and [35S]cysteine (10 μCi) in HS buffer containing 0.15 M NaCl in a final volume of 50 μl. After incubation at 30°C for 10 min, proteins were precipitated with 10% TCA. All samples were analysed by non-reducing SDS/PAGE followed by autoradiography. The intensity of Nfs1p-S–35SH in WT mitochondria in the absence of added nucleotides and iron (lane 1) was considered 100%. Arrowheads indicate bands (24 and 22 kDa) present in lanes 1, 3, 4 and 6. The asterisk indicates a major band (12 kDa) present in lanes 1 and 3. Molecular masses in kDa are indicated to the left-hand side.

To assess the persulfide-forming activity, mitochondria were depleted for endogenous nucleotides, and incubated with [35S]cysteine. Following removal of excess and free [35S]cysteine, samples were analysed by non-reducing SDS/PAGE and autoradiography. In WT mitochondria, a major radiolabelled protein (51 kDa; Nfs1p-S–35SH) was detected owing to the covalent [35S]persulfide formed on Nfs1p (Figure 1B, lane 1). The radiolabelled 51 kDa protein in mitochondria migrated slightly faster than the radiolabelled enzyme generated by incubating the purified Nfs1p·Isd11p complex with [35S]cysteine (Figure 1B, lane 7), because the Nfs1p enzyme of the purified complex contained a C-terminal His6 tag. Treatment with DTT prior to SDS/PAGE released the persulfide sulfur from Nfs1p, and no radiolabelled Nfs1p was detected (Supplementary Figure S2 at http://www.BiochemJ.org/bj/448/bj4480171add.htm) [26]. Mitochondria lacking Nfs1p (Nfs1p▼) served as a control, and, as expected, the 51 kDa radiolabelled band was not detected in these mitochondria (Figure 1B, lane 6). Several other radiolabelled proteins were also detected in both WT (Nfs1p present) and Nfs1p▼ (Nfs1p absent) mitochondria (Figure 1B, compare lanes 1 and 6). Radiolabelling of these bands was not due to incorporation of [35S]cysteine as an amino acid forming a covalent peptide bond in protein backbones, and these proteins lost the 35S signal after treatment with DTT (Supplementary Figure S2) [26]. Whereas some of these proteins in WT mitochondria might represent the recipients of persulfide sulfur from Nfs1p, the Nfs1p-independent background radiolabelling could be due to cysteinylation or H2S formation and S-sulfhydration of proteins [3436]. In any case, these results confirm our recent finding that isolated WT mitochondria are capable of forming new and radiolabelled persulfide on Nfs1p; the signal is specific for Nfs1p and can be distinguished from other radiolabelled background proteins by non-reducing SDS gels [26].

Isd11p▼ mitochondria contained Nfs1p (Figure 1A) and yet no Nfs1p-bound persulfide was detected (Figure 1B, lane 2). We therefore sought to determine whether Isd11p is really needed for the persulfide-forming activity of Nfs1p in mitochondria. For this purpose, Isd11p was expressed in bacteria and purified (Supplementary Figure S1A). Isd11p was incubated with isolated and Isd11p▼ mitochondria in a standard import reaction (Figure 1B, lanes 3 and 4; Isd11p import, 1st step). As a control, mitochondria were pretreated with valinomycin to dissipate membrane potential across the inner membrane, thereby blocking import (lane 5). Following incubation with Isd11p, mitochondria were recovered and incubated with [35S]cysteine (Figure 1B, [35S]cysteine, 2nd step) in the absence (lanes 3 and 5) or presence (lane 4) of added nucleotides and iron. Samples were then analysed by non-reducing SDS/PAGE followed by autoradiography.

Isd11p▼ mitochondria with imported Isd11p behaved just like WT mitochondria. Specifically, imported Isd11p restored Nfs1p persulfide formation in Isd11p▼ mitochondria to WT levels (Figure 1B, compare lanes 1 and 3). Such restoration was totally blocked when mitochondria were pretreated with valinomycin to inhibit Isd11p import (Figure 1B, lane 5). It is interesting to note that imported Isd11p also restored radiolabelling of some other proteins in Isd11p▼ mitochondria (Figure 1B, compare lanes 2 and 3). Two of these proteins (24 and 22 kDa, indicated by arrowheads) appeared to be background bands as they were also detected in Nfs1p▼ mitochondria (Figure 1B, compare lanes 3 and 6). By contrast, a 12 kDa protein (indicated by an asterisk) was found to be strongly radiolabelled in Isd11p▼ mitochondria with imported Isd11p (Figure 1B, lane 3) or WT mitochondria (Figure 1B, lane 1) but not in Nfs1p▼ mitochondria (Figure 1B, lane 6). The implication of this finding is addressed in the Discussion section.

We then asked whether Isd11p-mediated persulfide formation represents a bona fide biochemical intermediate for Fe–S cluster synthesis. Whereas the Nfs1p cysteine desulfurase activity does not require nucleotides or iron, Fe–S cluster synthesis in mitochondria is strictly dependent upon the optimum levels of nucleotides and available iron [1922,26]. In fact, addition of nucleotides and ferrous ascorbate led to a greatly reduced level of Nfs1p-S–35SH persulfide detected in Isd11p▼ mitochondria with imported Isd11p (Figure 1B, compare lanes 3 and 4), reflecting a balance between persulfide formation and persulfide transfer. Under these conditions, most probably the Nfs1p-bound persulfide was utilized for new Fe–35S cluster synthesis, and this issue is further addressed below.

Imported Isd11p was able to restore Nfs1p-bound persulfide in Isd11p▼ mitochondria. This raised the question of whether Isd11p is required for persulfide formation or for persulfide stabilization on Nfs1p. For example, the experiments described above could not rule out the possibility that in Isd11p▼ mitochondria, a covalent persulfide was generated on Nfs1p but was released by endogenous mitochondrial reductants. We therefore performed experiments using purified Nfs1p and Isd11p proteins in the absence of any other mitochondrial proteins, reductants or components. Specifically, Nfs1p–His6 and Isd11p–His6 were expressed separately in bacteria in soluble forms. After purification by Ni-NTA chromatography, both proteins were found to be almost 90% pure as judged by Coomassie Blue staining of SDS gels (Supplementary Figure S1A). The purified Nfs1p·Isd11p complex (formed in vivo in bacteria) served as a positive control for the Nfs1p cysteine desulfurase activity. As expected, upon incubation with [35S]cysteine, radiolabelled persulfide was formed on Nfs1p of the purified Nfs1p·Isd11p complex. By contrast, purified Nfs1p by itself was inactive. Nfs1p alone, incubated with [35S]cysteine, was not radiolabelled and hence failed to generate covalent persulfide (Figure 2A). More importantly, addition of purified Isd11p to purified Nfs1p resulted in strong radiolabelling of Nfs1p due to [35S]persulfide formation on the enzyme in a concentration-dependent manner (Figure 2B). No radiolabelled band was detected when Isd11p alone was incubated with [35S]cysteine. Thus Isd11p was required for persulfide formation. Most probably, purified Isd11p activated Nfs1p through direct interactions, allowing covalent persulfide formation on the enzyme.

Purified Nfs1p by itself is inactive and must be activated by Isd11p for cysteine desulfurase activity

Figure 2
Purified Nfs1p by itself is inactive and must be activated by Isd11p for cysteine desulfurase activity

(A) The purified Nfs1p·Isd11p complex (100 ng) or purified Nfs1p alone (100 ng) was incubated at 30°C for 10 min with PLP (0.15 mM) and [35S]cysteine (10 μCi) in HS buffer containing 0.15 M NaCl in a final volume of 50 μl. Proteins were precipitated with 10% TCA and analysed by non-reducing SDS/PAGE followed by autoradiography. (B) Different amounts of purified Isd11p (soluble and non-denatured) were added to purified Nfs1p, and the reaction mixtures were supplemented with PLP (0.15 mM) and [35S]cysteine (10 μCi). Following incubation at 30°C for 10 min, proteins were analysed as in (A) above. (C) The purified Nfs1p·Isd11p complex or purified Nfs1p alone was incubated with PLP, L-cysteine and DTT in capped tubes. After 1 h of incubation at 30°C, the reaction was stopped by the addition of NaOH. DPD and ferric chloride were then added, and the quantity of released sulfide in solution was assessed at 670 nm as described in the Materials and methods section [30]. Results of representative experiments are presented. Experiments were repeated several times. (D) The reaction mixtures contained different amounts of purified Nfs1p (0, 5 or 10 μg). As indicated, purified Isd11p (5 μg) was added from a concentrated stock of solubilized inclusion bodies (5 mg/ml in 50 mM Tris/HCl, pH 8.0, containing 8 M urea). PLP, L-cysteine and DTT were then added and sulfide formation was analysed as in (C) above. The urea concentration in the assay mixture ranged from 0 to 22 mM and had no effect on the measurements. Results are representative of several assays performed.

Figure 2
Purified Nfs1p by itself is inactive and must be activated by Isd11p for cysteine desulfurase activity

(A) The purified Nfs1p·Isd11p complex (100 ng) or purified Nfs1p alone (100 ng) was incubated at 30°C for 10 min with PLP (0.15 mM) and [35S]cysteine (10 μCi) in HS buffer containing 0.15 M NaCl in a final volume of 50 μl. Proteins were precipitated with 10% TCA and analysed by non-reducing SDS/PAGE followed by autoradiography. (B) Different amounts of purified Isd11p (soluble and non-denatured) were added to purified Nfs1p, and the reaction mixtures were supplemented with PLP (0.15 mM) and [35S]cysteine (10 μCi). Following incubation at 30°C for 10 min, proteins were analysed as in (A) above. (C) The purified Nfs1p·Isd11p complex or purified Nfs1p alone was incubated with PLP, L-cysteine and DTT in capped tubes. After 1 h of incubation at 30°C, the reaction was stopped by the addition of NaOH. DPD and ferric chloride were then added, and the quantity of released sulfide in solution was assessed at 670 nm as described in the Materials and methods section [30]. Results of representative experiments are presented. Experiments were repeated several times. (D) The reaction mixtures contained different amounts of purified Nfs1p (0, 5 or 10 μg). As indicated, purified Isd11p (5 μg) was added from a concentrated stock of solubilized inclusion bodies (5 mg/ml in 50 mM Tris/HCl, pH 8.0, containing 8 M urea). PLP, L-cysteine and DTT were then added and sulfide formation was analysed as in (C) above. The urea concentration in the assay mixture ranged from 0 to 22 mM and had no effect on the measurements. Results are representative of several assays performed.

We also measured the cysteine desulfurase activity of purified proteins using the colorimetric assay for sulfide production under reducing conditions. The purified Nfs1p·Isd11p complex was active, and activity was concentration-dependent. By contrast, Nfs1p alone was completely inactive within the limits of the assay (Figure 2C), and activity was observed only after addition of purified Isd11p (Figure 2D). No sulfide-forming activity was detected with Isd11p alone. In summary, radiolabelled persulfide was formed on Nfs1p of the Nfs1p·Isd11p complex but not on Nfs1p alone, correlating with the sulfide-forming spectrophotometric results and indicating a cysteine desulfurase activity of the complex and a lack of activity of the Nfs1p protein by itself. Addition of Isd11p to Nfs1p activated the enzyme, allowing persulfide formation on Nfs1p in the absence of DTT and sulfide formation in solution in the presence of DTT. In Isd11p▼ mitochondria, imported Isd11p probably interacted with Nfs1p, activating the enzyme and restoring the cysteine desulfurase activity.

Isd11p-independent substrate binding by Nfs1p followed by Isd11p-dependent covalent persulfide formation

The ability of imported Isd11p to ‘activate’ Nfs1p in Isd11p▼ mitochondria raises the question of how the effect is mediated. An assay was devised in which the substrate-binding step of the cysteine desulfurase could be separated from the persulfide-forming step. Mitochondria were incubated with [35S]cysteine (Figure 3A, 1st step), and excess [35S]cysteine was removed by washing. The recovered mitochondria were then ruptured by continuous bath sonication (FS14; Fisher Scientific), and the lysates were incubated with various amounts of purified Isd11p (Figure 3A, 2nd step). In the absence of added Isd11p, WT mitochondria exhibited Nfs1p persulfide and Isd11p▼ mitochondria did not (Figure 3A, lanes 1 and 3 respectively). Remarkably, addition of purified Isd11p to the Isd11p▼ lysate restored persulfide formation in a concentration-dependent manner (Figure 3A, lanes 4–6). These results can be interpreted as follows.

Nfs1p binds cysteine in the absence of Isd11p but persulfide formation from bound cysteine occurs only in the presence of Isd11p

Figure 3
Nfs1p binds cysteine in the absence of Isd11p but persulfide formation from bound cysteine occurs only in the presence of Isd11p

(A) Nucleotide-depleted mitochondria (WT, Isd11p▼ or Nfs1p▼) were incubated at 30°C for 15 min with 10 μCi of [35S]cysteine in HS buffer containing 10 mM magnesium acetate and 40 mM potassium acetate. Samples were diluted with HS buffer containing 0.15 M NaCl and left on ice for 10 min. After centrifugation at 15000 g for 5 min, the mitochondrial pellets were resuspended in 20 mM Hepes/KOH, pH 7.5. Mitochondrial membranes were ruptured by freezing the samples at −80°C followed by thawing and bath sonication for 30 s at 4°C. This process was repeated three times. Samples were then centrifuged at 15000 g for 15 min at 4°C, and the supernatant fractions were adjusted to 50 μl with final concentrations of 20 mM Hepes/KOH, pH 7.5, 10 mM magnesium acetate and 40 mM potassium acetate, with or without different amounts of soluble and non-denatured Isd11p. Following incubation at 30°C for 15 min, proteins were precipitated with the addition of equal volumes of ice-cold 20% TCA. Samples were centrifuged at 15000 g for 30 min at 4°C, and the pellets were analysed by non-reducing SDS/PAGE followed by autoradiography. The intensity of Nfs1p-S–35SH in WT mitochondrial lysate without any added Isd11p (lane 1) was considered 100%. The purified Nfs1p·Isd11p complex was incubated with [35S]cysteine and processed as in Figure 1(B), serving as a positive control for persulfide formation (lane 8). Arrowheads indicate bands (24 and 22 kDa) present in lanes 1, 2 and 4–7. The asterisk indicates a major band (12 kDa) present in lanes 1, 2 and 4–6. Molecular masses in kDa are indicated to the left-hand side. (B) Nfs1p (200 ng) was incubated at 30°C for 10 min with PLP (0.15 mM) and [35S]cysteine (10 μCi) in HS buffer containing 0.15 M NaCl in a final volume of 50 μl. Following the addition of saturated ammonium sulfate to a final concentration of 67%, samples were left on ice for 1 h, and then centrifuged at 15000 g for 30 min at 4°C. The protein precipitates thus obtained were dissolved in buffer A, and incubated at 30°C for 10 min, with or without added Isd11p (soluble and non-denatured) as indicated. Proteins were then precipitated with 10% TCA and analysed by non-reducing SDS/PAGE followed by autoradiography.

Figure 3
Nfs1p binds cysteine in the absence of Isd11p but persulfide formation from bound cysteine occurs only in the presence of Isd11p

(A) Nucleotide-depleted mitochondria (WT, Isd11p▼ or Nfs1p▼) were incubated at 30°C for 15 min with 10 μCi of [35S]cysteine in HS buffer containing 10 mM magnesium acetate and 40 mM potassium acetate. Samples were diluted with HS buffer containing 0.15 M NaCl and left on ice for 10 min. After centrifugation at 15000 g for 5 min, the mitochondrial pellets were resuspended in 20 mM Hepes/KOH, pH 7.5. Mitochondrial membranes were ruptured by freezing the samples at −80°C followed by thawing and bath sonication for 30 s at 4°C. This process was repeated three times. Samples were then centrifuged at 15000 g for 15 min at 4°C, and the supernatant fractions were adjusted to 50 μl with final concentrations of 20 mM Hepes/KOH, pH 7.5, 10 mM magnesium acetate and 40 mM potassium acetate, with or without different amounts of soluble and non-denatured Isd11p. Following incubation at 30°C for 15 min, proteins were precipitated with the addition of equal volumes of ice-cold 20% TCA. Samples were centrifuged at 15000 g for 30 min at 4°C, and the pellets were analysed by non-reducing SDS/PAGE followed by autoradiography. The intensity of Nfs1p-S–35SH in WT mitochondrial lysate without any added Isd11p (lane 1) was considered 100%. The purified Nfs1p·Isd11p complex was incubated with [35S]cysteine and processed as in Figure 1(B), serving as a positive control for persulfide formation (lane 8). Arrowheads indicate bands (24 and 22 kDa) present in lanes 1, 2 and 4–7. The asterisk indicates a major band (12 kDa) present in lanes 1, 2 and 4–6. Molecular masses in kDa are indicated to the left-hand side. (B) Nfs1p (200 ng) was incubated at 30°C for 10 min with PLP (0.15 mM) and [35S]cysteine (10 μCi) in HS buffer containing 0.15 M NaCl in a final volume of 50 μl. Following the addition of saturated ammonium sulfate to a final concentration of 67%, samples were left on ice for 1 h, and then centrifuged at 15000 g for 30 min at 4°C. The protein precipitates thus obtained were dissolved in buffer A, and incubated at 30°C for 10 min, with or without added Isd11p (soluble and non-denatured) as indicated. Proteins were then precipitated with 10% TCA and analysed by non-reducing SDS/PAGE followed by autoradiography.

The [35S]cysteine added to Isd11p▼ mitochondria during the 1st step bound to Nfs1p through interactions with PLP in the substrate-binding site (Isd11p-independent step). This form of Nfs1p was not detected by SDS gels because SDS released the radioactive substrate from the [35S]cysteine–PLP ketimine adduct. The purified Isd11p added to the lysate during the 2nd step induced Nfs1p with prebound [35S]cysteine to form the covalent Nfs1p-S–35SH persulfide (Isd11p-dependent step). The covalently bound persulfide resisted SDS and was detected by the non-reducing gel system. Thus Isd11p was not required for Nfs1p binding to the substrate cysteine in mitochondria but it was essential for persulfide formation at the active site cysteine residue from the bound substrate cysteine. Most probably, Isd11p binding induced an activating conformational change in Nfs1p. This idea is consistent with the data of similar experiments performed with purified proteins (Figure 3B). Briefly, purified Nfs1p was incubated with [35S]cysteine. Following removal of excess and unbound radioactive substrate, Nfs1p with bound [35S]cysteine was incubated with different amounts of purified Isd11p. Samples were analysed by non-reducing SDS/PAGE followed by autoradiography. A covalent persulfide on Nfs1p (Nfs1p-S–35SH) was detected only when Isd11p was added to Nfs1p with the non-covalently bound substrate [35S]cysteine (Figure 3B). These results suggest that a direct interaction of Isd11p with Nfs1p is sufficient for inducing an activating conformational change in the enzyme, although other proteins may also be involved in the activation process in mitochondria.

NEM sensitivity of the Nfs1p·Isd11p complex compared with Nfs1p alone

Isd11p does not contain any cysteine residues, and therefore it is unlikely to be affected by treatment with alkylating agents. On the other hand, cysteine desulfurases are generally sensitive to alkylating agents because of the critical role of the active site cysteine residue in the enzymatic reaction [12]. For example, A. vinelandii NifS is exquisitely sensitive to NEM, being almost completely inactivated at concentrations of approximately 0.1 mM [26]. In a set of experiments, WT mitochondria containing the Nfs1p·Isd11p complex or Isd11p▼ mitochondria containing Nfs1p alone were treated with the membrane-permeable alkylator NEM followed by evaluation of the Nfs1p persulfide-forming activity (Figure 4). The experimental design involved treatment of mitochondria with the alkylating agent NEM (1 mM) followed by neutralization with DTT (10 mM). A sample in which the order of addition was reversed, and DTT was added prior to NEM treatment, was included as a negative control. For assessment of persulfide-forming activity, mitochondria were incubated with [35S]cysteine, recovered, and analysed by non-reducing SDS gels followed by autoradiography (Figure 4). In some samples with Isd11p▼ mitochondria, Isd11p was imported prior to incubation with [35S]cysteine.

NEM sensitivity of the active site cysteine residue of Nfs1p in the absence or presence of Isd11p in mitochondria

Figure 4
NEM sensitivity of the active site cysteine residue of Nfs1p in the absence or presence of Isd11p in mitochondria

Mitochondria (200 μg of protein) were incubated with NEM (1 mM) for 10 min at 30°C. Following addition of DTT (10 mM), samples were again incubated for 10 min at 30°C to neutralize free NEM. These samples are marked ‘NEM/DTT’. Some samples were treated with DTT (10 mM) first at 30°C for 10 min. NEM (1 mM) was then added and incubation at 30°C was continued for another 10 min. These samples are marked ‘DTT/NEM’. Some other samples were incubated at 30°C for 20 min with DTT alone (marked ‘DTT’), or with no DTT or NEM (marked ‘Control pretreatment’). Reaction mixtures were diluted with HS buffer and mitochondria were recovered by centrifugation at 15000 g for 5 min at 4°C. Isd11p▼ mitochondria were then incubated with or without urea-denatured and unlabelled Isd11p (100 ng) for 15 min at 30°C under import conditions as in Figure 1(B). Samples were diluted with HS buffer, and mitochondria, with or without imported Isd11p, were recovered by centrifugation. Isd11p▼ mitochondria pretreated with valinomycin served as a negative control for the import reaction. All mitochondrial samples were then incubated at 30°C for 15 min with 10 μCi of [35S]cysteine in HS buffer containing 10 mM magnesium acetate and 40 mM potassium acetate, and persulfide formation was evaluated as in Figure 1(B). The intensity of Nfs1p-S–35SH in WT mitochondria with control pretreatment (lane 1) was considered 100%. The purified Nfs1p·Isd11p complex incubated with [35S]cysteine served as a positive control for Nfs1p-bound persulfide (lane 11). Arrowheads indicate bands (24 and 22 kDa) present in lanes 1–3, 5, 6, 8 and 10. The asterisk indicates a major band (12 kDa) present in lanes 1–3, 5, 6 and 8. Molecular masses in kDa are indicated to the left-hand side.

Figure 4
NEM sensitivity of the active site cysteine residue of Nfs1p in the absence or presence of Isd11p in mitochondria

Mitochondria (200 μg of protein) were incubated with NEM (1 mM) for 10 min at 30°C. Following addition of DTT (10 mM), samples were again incubated for 10 min at 30°C to neutralize free NEM. These samples are marked ‘NEM/DTT’. Some samples were treated with DTT (10 mM) first at 30°C for 10 min. NEM (1 mM) was then added and incubation at 30°C was continued for another 10 min. These samples are marked ‘DTT/NEM’. Some other samples were incubated at 30°C for 20 min with DTT alone (marked ‘DTT’), or with no DTT or NEM (marked ‘Control pretreatment’). Reaction mixtures were diluted with HS buffer and mitochondria were recovered by centrifugation at 15000 g for 5 min at 4°C. Isd11p▼ mitochondria were then incubated with or without urea-denatured and unlabelled Isd11p (100 ng) for 15 min at 30°C under import conditions as in Figure 1(B). Samples were diluted with HS buffer, and mitochondria, with or without imported Isd11p, were recovered by centrifugation. Isd11p▼ mitochondria pretreated with valinomycin served as a negative control for the import reaction. All mitochondrial samples were then incubated at 30°C for 15 min with 10 μCi of [35S]cysteine in HS buffer containing 10 mM magnesium acetate and 40 mM potassium acetate, and persulfide formation was evaluated as in Figure 1(B). The intensity of Nfs1p-S–35SH in WT mitochondria with control pretreatment (lane 1) was considered 100%. The purified Nfs1p·Isd11p complex incubated with [35S]cysteine served as a positive control for Nfs1p-bound persulfide (lane 11). Arrowheads indicate bands (24 and 22 kDa) present in lanes 1–3, 5, 6, 8 and 10. The asterisk indicates a major band (12 kDa) present in lanes 1–3, 5, 6 and 8. Molecular masses in kDa are indicated to the left-hand side.

As noted for Figure 1(B), Isd11p▼ mitochondria were unable to form the Nfs1p-S–35SH persulfide (Figure 4, lane 4), but they recovered the activity when Isd11p was imported into these mitochondria (Figure 4, lanes 5 and 6). The persulfide-forming activity was not recovered when Isd11p▼ mitochondria (containing uncomplexed Nfs1p) were treated with NEM followed by neutralization with DTT and import of Isd11p (lane 7). By contrast, WT mitochondria (containing the preformed and endogenous Nfs1p·Isd11p complex) mostly maintained their persulfide-forming activity after treatment with NEM followed by neutralization with DTT (lane 2). This striking difference probably indicates that the preformed Nfs1p·Isd11p complex is more resistant to the alkylator than the uncomplexed Nfs1p. These results also suggest that Isd11p binding to Nfs1p induces a conformational change that protects the active site cysteine residue from being modified by NEM. Key controls for these experiments are shown in Supplementary Figure S3 (at http://www.BiochemJ.org/bj/448/bj4480171add.htm). NEM pretreatment of Isd11p▼ mitochondria at a concentration that abrogated activation by Isd11p did not interfere with Isd11p import (Supplementary Figure S3, compare lanes 7 and 9). Restoration of persulfide-forming activity to Isd11p▼ mitochondria was abolished by inclusion of valinomycin (Figure 4, lane 9) because, as expected, Isd11p import was completely blocked under these conditions (Supplementary Figure S3, lane 11).

To validate further the results of the NEM sensitivity of Nfs1p in mitochondria, we performed a similar two-step experiment with the purified proteins (Figure 5). Briefly, the Nfs1p·Isd11p complex or Nfs1p by itself was treated with increasing concentrations of NEM, and then excess NEM was neutralized with DTT (Figure 5, 1,st step). Proteins were precipitated with ammonium sulfate to remove NEM and DTT and resuspended in buffer. Isd11p was added to reaction mixtures containing NEM-treated or -untreated Nfs1p alone, but not to the corresponding Nfs1p·Isd11p samples (Figure 5, 2,nd step). Following incubation with [35S]cysteine, radiolabelled persulfide formation on Nfs1p was evaluated. Nfs1p by itself was found to be much more sensitive to NEM-mediated inactivation than the Nfs1p·Isd11p complex. For example, Nfs1p alone treated with 1 mM NEM followed by supplementation with Isd11p completely failed to generate persulfide (Figure 5, lane 10). By contrast, pretreatment of the preformed Nfs1p·Isd11p complex with 1 mM NEM only slightly affected persulfide formation (Figure 5, lane 4), and some persulfide formation occurred even after pretreatment with a 4-fold higher concentration of NEM (Figure 5, lane 5). These results are in good agreement with the enhanced NEM sensitivity of Nfs1p in Isd11p▼ mitochondria (Figure 4) and further substantiate the notion that Isd11p activates the enzyme activity of Nfs1p while protecting the active site cysteine residue from NEM inactivation. An Isd11p-mediated conformational change in Nfs1p would be in agreement with this idea.

NEM sensitivity of the active site cysteine residue of the purified Nfs1p·Isd11p complex compared with Nfs1p alone

Figure 5
NEM sensitivity of the active site cysteine residue of the purified Nfs1p·Isd11p complex compared with Nfs1p alone

The preformed Nfs1p·Isd11p complex (100 ng) or Nfs1p by itself (100 ng) was treated with increasing concentrations of NEM for 10 min at 30°C. DTT (10 mM) was added and incubation at 30°C was continued for another 5 min to neutralize free NEM. Proteins were precipitated with ammonium sulfate as in Figure 3(B), and the pellets were solubilized with buffer A. Purified Isd11p (soluble and non-denatured; 100 ng) was then added only to Nfs1p alone samples as indicated. All reaction mixtures were supplemented with PLP (0.15 mM) and [35S]cysteine (10 μCi), and incubated at 30°C for 10 min. After precipitation with 10% TCA, the protein pellets were analysed by non-reducing SDS/PAGE followed by autoradiography. The intensity of Nfs1p-S–35SH for NEM-untreated Nfs1p·Isd11p (lane 1) was considered 100%.

Figure 5
NEM sensitivity of the active site cysteine residue of the purified Nfs1p·Isd11p complex compared with Nfs1p alone

The preformed Nfs1p·Isd11p complex (100 ng) or Nfs1p by itself (100 ng) was treated with increasing concentrations of NEM for 10 min at 30°C. DTT (10 mM) was added and incubation at 30°C was continued for another 5 min to neutralize free NEM. Proteins were precipitated with ammonium sulfate as in Figure 3(B), and the pellets were solubilized with buffer A. Purified Isd11p (soluble and non-denatured; 100 ng) was then added only to Nfs1p alone samples as indicated. All reaction mixtures were supplemented with PLP (0.15 mM) and [35S]cysteine (10 μCi), and incubated at 30°C for 10 min. After precipitation with 10% TCA, the protein pellets were analysed by non-reducing SDS/PAGE followed by autoradiography. The intensity of Nfs1p-S–35SH for NEM-untreated Nfs1p·Isd11p (lane 1) was considered 100%.

Hypomorphic Isd11p mutant allele confers defects in mitochondrial Fe–S clusters and iron homoeostasis

Isd11p is conserved throughout eukaryotes, and it shares a few primary amino acid sequence elements with homologues from other species. There is also a larger family of proteins characterized by the presence of the tripeptide motif LYR/K. Members of this family include accessory subunits B22 and B14 of the mitochondrial complex I [23], the Sdh6p protein involved in the assembly of complex II [37], and the Mzm1p protein implicated in a late stage of complex III assembly [38]. To test the function of the motif L15YK17 of Isd11p, it was mutated to A15AA17 in a centromere-based plasmid carrying authentic Isd11p coding and flanking sequences. The pRS416 empty plasmid, pRS416 with authentic Isd11p and pRS416 with Isd11p mut (Isd11p L15YK17 changed to A15AA17) were introduced into the Gal-Isd11p strain, and the galactose-regulated promoter was repressed by spotting on to glucose agar plates, thereby allowing evaluation of plasmid-dependent growth. The empty plasmid conferred very little growth, consistent with the essential nature of Isd11p. Whereas the authentic Isd11p restored normal growth, the Isd11p mut conferred slower growth, consistent with a partial loss of function (Figure 6A, right-hand panel). In fact, the Isd11p mut was associated with phenotypes characteristic of Fe–S cluster assembly deficiency. Fe–S cluster enzyme activities of aconitase and succinate dehydrogenase in mitochondria were severely depleted. Malate dehydrogenase does not contain any Fe–S clusters and its activity was only slightly affected, thereby serving as a control (Figure 6B). Furthermore, iron accumulated in the mutant cells and mutant mitochondria due to dysregulation of iron homoeostasis is characteristic of Fe–S cluster deficiency (Figure 6C). These results suggest that the L15YK17 to A15AA17 Isd11p mutant reduces, but does not completely abrogate, Isd11p function in Fe–S cluster biogenesis.

Effects of an Isd11p mutation on cell growth, Fe–S cluster enzyme activities and iron homoeostasis

Figure 6
Effects of an Isd11p mutation on cell growth, Fe–S cluster enzyme activities and iron homoeostasis

(A) Serial 1:5 dilutions starting with 105 cells of Gal-Isd11p transformants with pRS416 (top), pRS416/Isd11p (middle) or pRS416/Isd11p mut (bottom) were spotted on to agar plates containing defined medium minus uracil with 2% raffinose and 0.5% galactose (left-hand panel, Gal-Isd11p▲) or 2% glucose (right-hand panel, Gal-Isd11p▼) as the carbon source. The plates were incubated for 4 days at 30°C and photographed. Prior to spotting, each transformant was grown in defined raffinose-galactose medium minus uracil to logarithmic phase and washed free of growth medium. (B) The Gal-Isd11p transformants carrying pRS416/Isd11p or pRS416/Isd11p mut were grown in defined minus uracil medium with raffinose as the carbon source (Gal-Isd11p▼). Mitochondria were isolated and the enzyme activities were measured for aconitase (ACO), succinate dehydrogenase (SDH) and malate dehydrogenase (MDH). For each enzyme, the results are reported as a ratio of the activity of the mutant (pRS416/Isd11p mut) mitochondria compared with the activity of the control (pRS416/Isd11p) mitochondria. (C) The Gal-Isd11p transformants carrying pRS416/Isd11p or pRS416/Isd11p mut were grown for 14 h in defined minus uracil medium with raffinose as the carbon source and supplemented with 100 nM 55Fe ferrous ascorbate. The cells were recovered and washed with 50 mM Tris/HCl, pH 7.5, containing 5 mM EDTA. Mitochondria were isolated and washed with 50 mM Tris/HCl, pH 7.5, containing 0.6 M sorbitol and 5 mM EDTA. Iron was quantified by scintillation counting for total cellular iron (103 c.p.m. per 100 D600 cells, bars 1 and 2) and for mitochondrial iron (103 c.p.m. per 10 mg of mitochondrial proteins, bars 3 and 4). The results for control (Isd11p-expressing) and mutant (Isd11p mut) cells are depicted by the open and closed bars respectively.

Figure 6
Effects of an Isd11p mutation on cell growth, Fe–S cluster enzyme activities and iron homoeostasis

(A) Serial 1:5 dilutions starting with 105 cells of Gal-Isd11p transformants with pRS416 (top), pRS416/Isd11p (middle) or pRS416/Isd11p mut (bottom) were spotted on to agar plates containing defined medium minus uracil with 2% raffinose and 0.5% galactose (left-hand panel, Gal-Isd11p▲) or 2% glucose (right-hand panel, Gal-Isd11p▼) as the carbon source. The plates were incubated for 4 days at 30°C and photographed. Prior to spotting, each transformant was grown in defined raffinose-galactose medium minus uracil to logarithmic phase and washed free of growth medium. (B) The Gal-Isd11p transformants carrying pRS416/Isd11p or pRS416/Isd11p mut were grown in defined minus uracil medium with raffinose as the carbon source (Gal-Isd11p▼). Mitochondria were isolated and the enzyme activities were measured for aconitase (ACO), succinate dehydrogenase (SDH) and malate dehydrogenase (MDH). For each enzyme, the results are reported as a ratio of the activity of the mutant (pRS416/Isd11p mut) mitochondria compared with the activity of the control (pRS416/Isd11p) mitochondria. (C) The Gal-Isd11p transformants carrying pRS416/Isd11p or pRS416/Isd11p mut were grown for 14 h in defined minus uracil medium with raffinose as the carbon source and supplemented with 100 nM 55Fe ferrous ascorbate. The cells were recovered and washed with 50 mM Tris/HCl, pH 7.5, containing 5 mM EDTA. Mitochondria were isolated and washed with 50 mM Tris/HCl, pH 7.5, containing 0.6 M sorbitol and 5 mM EDTA. Iron was quantified by scintillation counting for total cellular iron (103 c.p.m. per 100 D600 cells, bars 1 and 2) and for mitochondrial iron (103 c.p.m. per 10 mg of mitochondrial proteins, bars 3 and 4). The results for control (Isd11p-expressing) and mutant (Isd11p mut) cells are depicted by the open and closed bars respectively.

Isolated Isd11p mutant mitochondria lack persulfide-forming activity that is restored by imported authentic Isd11p

Mitochondria isolated from the Gal-Isd11p transformants expressing Isd11p mut were evaluated for the Nfs1p-S–35SH persulfide formation. As expected, mitochondria with authentic Isd11p (WT) formed radiolabelled persulfide (Figure 7, lane 1). However, mitochondria containing the Isd11p mut protein were inactive (Figure 7, lane 3). More importantly, import of authentic Isd11p into the Isd11p mut mitochondria restored persulfide formation. The activity correction was dependent upon the concentration of Isd11p used for import (Figure 7, lanes 4 and 5) and was blocked by pretreatment of mitochondria with valinomycin (Figure 7, lane 6). These results imply that the Isd11p mut (A15AA17) protein expressed in vivo was unable to activate the Nfs1p protein to form persulfide in isolated mitochondria and that the authentic Isd11p (L15YK17) protein imported into these mitochondria restored activity. The imported authentic Isd11p probably displaced the Isd11p mut protein from the inactive Nfs1p·Isd11p mut complex as part of the activation process. The affinities of the authentic and mutant Isd11p proteins for interaction with Nfs1p remain to be determined.

Import of authentic Isd11p restores persulfide formation in isolated Isd11p mutant mitochondria

Figure 7
Import of authentic Isd11p restores persulfide formation in isolated Isd11p mutant mitochondria

Mitochondria were isolated from Isd11p▼ (Gal-Isd11p▼) cells transformed with pRS416/Isd11p mut. Nfs1p▼ mitochondria served as a negative control. Where indicated, a standard import reaction was performed to introduce bacterially expressed and urea-denatured Isd11p into WT or Isd11p mut mitochondria as described in Figure 1(B) (1st step). After the import reaction, mitochondria were recovered and incubated with [35S]cysteine for 15 min at 30°C in the absence of added nucleotides or iron (2nd step). Radiolabelled persulfide was detected by non-reducing SDS/PAGE followed by autoradiography. In one sample shown in lane 6, mitochondria were pretreated with valinomycin to dissipate the membrane potential prior to the import reaction. As a positive control (lane 8), the purified Nfs1p·Isd11p complex was incubated with [35S]cysteine and processed as in Figure 1(B). Arrowheads indicate bands (24 and 22 kDa) present in lanes 1, 2, 4, 5 and 7. The asterisk indicates a major band (12 kDa) present in lanes 1, 2, 4 and 5. Molecular masses in kDa are indicated to the left-hand side.

Figure 7
Import of authentic Isd11p restores persulfide formation in isolated Isd11p mutant mitochondria

Mitochondria were isolated from Isd11p▼ (Gal-Isd11p▼) cells transformed with pRS416/Isd11p mut. Nfs1p▼ mitochondria served as a negative control. Where indicated, a standard import reaction was performed to introduce bacterially expressed and urea-denatured Isd11p into WT or Isd11p mut mitochondria as described in Figure 1(B) (1st step). After the import reaction, mitochondria were recovered and incubated with [35S]cysteine for 15 min at 30°C in the absence of added nucleotides or iron (2nd step). Radiolabelled persulfide was detected by non-reducing SDS/PAGE followed by autoradiography. In one sample shown in lane 6, mitochondria were pretreated with valinomycin to dissipate the membrane potential prior to the import reaction. As a positive control (lane 8), the purified Nfs1p·Isd11p complex was incubated with [35S]cysteine and processed as in Figure 1(B). Arrowheads indicate bands (24 and 22 kDa) present in lanes 1, 2, 4, 5 and 7. The asterisk indicates a major band (12 kDa) present in lanes 1, 2, 4 and 5. Molecular masses in kDa are indicated to the left-hand side.

Isolated Isd11p▼ or Isd11p mut mitochondria lack Fe–S cluster biosynthesis activity that is restored by imported Isd11p

The Isd11p▼ (Figure 1B) or Isd11p mut (Figure 7) mitochondria were deficient in Nfs1p persulfide-forming activity and regained this activity following import of purified Isd11p. We therefore determined whether the restored persulfide-forming activity was sufficient for new Fe–S cluster formation on endogenous apoaconitase. For this purpose, purified Isd11p was imported into Isd11p▼ or Isd11p mut mitochondria in the presence of [35S]cysteine with added nucleotides (ATP, GTP and NADH), and ferrous ascorbate as the iron source. Samples were analysed by native PAGE followed by autoradiography. WT mitochondria were used as the positive control and showed efficient Fe–35S cluster labelling of aconitase (Aco1p) that increased over time (Figure 8, lanes 1 and 4). More importantly, following the import of purified Isd11p, both Isd11p▼ (Figure 8, lanes 2 and 5) and Isd11p mut (Figure 8, lanes 3 and 6) mitochondria showed efficient Fe–S cluster formation practically indistinguishable from WT mitochondria. As negative controls, when Isd11p▼ (Figure 8, lane 7) or Isd11p mut (Figure 8, lane 8) mitochondria were pretreated with valinomycin to block Isd11p import, no radiolabelled aconitase was detected. The imported Isd11p thus restored not only the persulfide-forming activity (Figures 1B, 4 and 7) but also the Fe–S cluster-forming activity to the Isd11p▼ or Isd11p mut mitochondria (Figure 8). We conclude that Isd11p activates the cysteine desulfurase activity of Nfs1p, thereby allowing formation of a productive persulfide at the active site cysteine of the enzyme for subsequent Fe–S cluster synthesis in mitochondria. In the absence of Isd11p, Nfs1p may bind the substrate cysteine but cannot form the persulfide. Consequently, Isd11p▼ or Isd11p mut mitochondria are deficient in Fe–S cluster formation.

Imported Isd11p restores Fe–S cluster biogenesis in Isd11p▼ mitochondria or in mitochondria with a mutant form of Isd11p

Figure 8
Imported Isd11p restores Fe–S cluster biogenesis in Isd11p▼ mitochondria or in mitochondria with a mutant form of Isd11p

All reaction mixtures contained mitochondria (200 μg of protein) in HS buffer with 10 mM magnesium acetate, 40 mM potassium acetate, 4 mM ATP, 1 mM GTP, 2 mM NADH, 10 μM ferrous ascorbate and 10 μCi of [35S]cysteine. In some cases, mitochondria were pretreated with valinomycin (5 μM) for 5 min on ice. Urea-denatured Isd11p (100 ng) was added to samples as indicated. The reaction volume was 50 μl, and the final urea concentration was 160 mM. After incubation at 30°C for 30–60 min, samples were diluted with HS buffer, and mitochondria were recovered. Mitochondrial membranes were ruptured, and soluble proteins were treated with DTT (10 mM) and analysed by native PAGE followed by autoradiography [1922,26].

Figure 8
Imported Isd11p restores Fe–S cluster biogenesis in Isd11p▼ mitochondria or in mitochondria with a mutant form of Isd11p

All reaction mixtures contained mitochondria (200 μg of protein) in HS buffer with 10 mM magnesium acetate, 40 mM potassium acetate, 4 mM ATP, 1 mM GTP, 2 mM NADH, 10 μM ferrous ascorbate and 10 μCi of [35S]cysteine. In some cases, mitochondria were pretreated with valinomycin (5 μM) for 5 min on ice. Urea-denatured Isd11p (100 ng) was added to samples as indicated. The reaction volume was 50 μl, and the final urea concentration was 160 mM. After incubation at 30°C for 30–60 min, samples were diluted with HS buffer, and mitochondria were recovered. Mitochondrial membranes were ruptured, and soluble proteins were treated with DTT (10 mM) and analysed by native PAGE followed by autoradiography [1922,26].

DISCUSSION

Fe–S cluster biogenesis in yeast mitochondria requires many proteins, including a cysteine desulfurase, Nfs1p, and an accessory protein, Isd11p [4,5,14]. The enzyme Nfs1p acts on the substrate cysteine, removing sulfur and forming a covalent persulfide at the active site cysteine. The persulfide sulfur is then transferred to the Isu1p/Isu2p scaffold and assembled with iron to form an Fe–S cluster intermediate in a step requiring the yeast frataxin homologue Yfh1p (Figure 9). Finally, the Fe–S cluster intermediate is delivered to apoproteins such as aconitase in a chaperone-dependent manner, forming holo or active proteins. Whereas Nfs1p, scaffold proteins, frataxin and chaperones are conserved from bacteria to humans, Isd11p is only found in eukaryotes and not in prokaryotes [23,24]. In the present study we show for the first time that Isd11p is absolutely required for the cysteine desulfurase activity of Nfs1p in mitochondria. In the absence of Isd11p in mitochondria, Nfs1p can bind the substrate cysteine but cannot form the covalent persulfide at the active site cysteine residue. The enzyme must be activated through interactions with Isd11p for persulfide generation. The results of the present study also show that Isd11p-mediated formation of Nfs1p-bound persulfide represents an authentic biochemical intermediate for Fe–S cluster synthesis in mitochondria. The genuine mitochondrial cysteine desulfurase activity requires both Nfs1p and Isd11p and is therefore different from the prokaryotic enzymes, which are active without any accessory proteins.

A model for Isd11p-mediated activation of Nfs1p and Fe–S cluster biogenesis of aconitase in mitochondria

Figure 9
A model for Isd11p-mediated activation of Nfs1p and Fe–S cluster biogenesis of aconitase in mitochondria

A demonstration of some of the key steps involved in Fe–S cluster biogenesis in mitochondria. Step 1: the substrate cysteine binds to Nfs1p through the cofactor PLP associated with a conserved Lys263 of the Nfs1p substrate-binding site, forming a cysteine–PLP ketimine adduct. In the absence of Isd11p, Nfs1p by itself cannot generate a covalent persulfide on the active site cysteine residue from the bound substrate and consequently the subsequent steps do not occur. Step 2: Isd11p binds to Nfs1p and forms a complex (Nfs1p·Isd11p). Such a process leads to an activating conformational change in the enzyme, bringing Cys385 of the active site into proximity with the substrate cysteine bound to the substrate-binding site. As part of the induced conformational change, the active site Cys385 becomes more resistant to alkylating agents. Step 3: a nucleophilic attack by the thiolate anion of the active site Cys385 extracts the -SH group of the substrate from the ketimine adduct, forming a covalent persulfide (Nfs1p-S–SH). The substrate cysteine is converted into alanine, which is released from the substrate-binding site, allowing rebinding of PLP and initiation of a new catalytic cycle. Step 4: the persulfide sulfur is transferred from Nfs1p to the Isu1p scaffold protein and is assembled with iron to form [Fe–S] cluster intermediates in a process requiring Yfh1p, the yeast frataxin homologue. Step 5: finally, the cluster intermediates are delivered from the scaffold protein to apoproteins such as aconitase in a chaperone-dependent manner. The PLP-conjugating Lys263 and the active site Cys385 of Nfs1p correspond to the mature form of the protein in mitochondria [15,18]. In addition to the proteins shown here, several other proteins are also involved in Fe–S cluster assembly in mitochondria [4,5,14].

Figure 9
A model for Isd11p-mediated activation of Nfs1p and Fe–S cluster biogenesis of aconitase in mitochondria

A demonstration of some of the key steps involved in Fe–S cluster biogenesis in mitochondria. Step 1: the substrate cysteine binds to Nfs1p through the cofactor PLP associated with a conserved Lys263 of the Nfs1p substrate-binding site, forming a cysteine–PLP ketimine adduct. In the absence of Isd11p, Nfs1p by itself cannot generate a covalent persulfide on the active site cysteine residue from the bound substrate and consequently the subsequent steps do not occur. Step 2: Isd11p binds to Nfs1p and forms a complex (Nfs1p·Isd11p). Such a process leads to an activating conformational change in the enzyme, bringing Cys385 of the active site into proximity with the substrate cysteine bound to the substrate-binding site. As part of the induced conformational change, the active site Cys385 becomes more resistant to alkylating agents. Step 3: a nucleophilic attack by the thiolate anion of the active site Cys385 extracts the -SH group of the substrate from the ketimine adduct, forming a covalent persulfide (Nfs1p-S–SH). The substrate cysteine is converted into alanine, which is released from the substrate-binding site, allowing rebinding of PLP and initiation of a new catalytic cycle. Step 4: the persulfide sulfur is transferred from Nfs1p to the Isu1p scaffold protein and is assembled with iron to form [Fe–S] cluster intermediates in a process requiring Yfh1p, the yeast frataxin homologue. Step 5: finally, the cluster intermediates are delivered from the scaffold protein to apoproteins such as aconitase in a chaperone-dependent manner. The PLP-conjugating Lys263 and the active site Cys385 of Nfs1p correspond to the mature form of the protein in mitochondria [15,18]. In addition to the proteins shown here, several other proteins are also involved in Fe–S cluster assembly in mitochondria [4,5,14].

The requirement of Isd11p for Nfs1p persulfide formation was in addition to and separate from its role in stabilizing Nfs1p in mitochondria. When Isd11p expression from the regulated GAL1 promoter was shut down, the Isd11p protein level rapidly declined, becoming almost undetectable within 12 h. Isd11p depletion also led to a reduction in Nfs1p protein level. However, a time window was identified during which Isd11p was severely depleted but Nfs1p persisted in mitochondria (Figure 1A). This allowed us to determine the enzyme activity in isolated mitochondria in the absence of Isd11p. Mitochondria isolated during this window lacked persulfide-forming activity and failed to generate Nfs1p-S–35SH when incubated with the radiolabelled substrate of Nfs1p, [35S]cysteine. However, persulfide formation was restored by the import of bacterially expressed and purified Isd11p (Figure 1B). The inability of Isd11p▼ mitochondria to generate the Nfs1p persulfide was most probably due to the lack of Isd11p and not due to secondary effects that might have occurred during in vivo depletion of Isd11p. Such a notion is strongly supported by the experiments with purified proteins. Radiolabelled persulfide was formed on Nfs1p of the purified Nfs1p·Isd11p complex (formed in vivo in bacteria). By contrast, purified Nfs1p was inactive on its own, and the enzyme was able to form the persulfide only after supplementation with purified Isd11p. These observations correlated well with spectrophotometric results for sulfide-forming cysteine desulfurase activity or lack of activity of the corresponding samples (Figure 2).

Isd11p imported into Isd11p▼ mitochondria resulted in strong persulfide formation on Nfs1p, particularly under nucleotide-depleted conditions that did not permit Fe–S cluster assembly. The persulfide could be chased by nucleotide and iron addition (Figure 1B), reflecting persulfide transfer to designated recipients in mitochondria. In fact, imported Isd11p restored not only persulfide formation but also Fe–S cluster synthesis in Isd11p-deficient mitochondria (Figure 8). Isd11p-mediated persulfide formation on Nfs1p therefore represents a genuine biochemical intermediate for Fe–S cluster synthesis in mitochondria. An Isd11p mutant with the conserved tripeptide L15YK17 motif changed to A15AA17 was examined, and this mutant was able to support slow growth (Figure 6A). Mitochondria isolated from the mutant strain failed to form persulfide and to synthesize Fe–S clusters but regained these activities upon the import of bacterially expressed and purified authentic Isd11p (Figures 7 and 8 respectively). Thus the LYK motif was necessary for the Nfs1p-activating function of the accessory protein.

Cysteine desulfurases require PLP bound as the cofactor at a conserved lysine residue in the substrate-binding site. In the enzyme reaction cycle, the substrate cysteine binds to PLP, forming a cysteine–PLP ketimine adduct. The bound substrate cysteine is then subjected to a nucleophilic attack by a cysteine residue in the active site loop of the enzyme, forming a persulfide adduct on this active site cysteine residue [8,13,39]. In the present study we have shown experimentally that these discrete stages of the enzyme reaction in mitochondria can be separated. The substrate binding to Nfs1p occurred first, followed by formation of the persulfide on the enzyme in an Isd11p-dependent manner. Mitochondria with Nfs1p and no Isd11p allowed binding of the substrate [35S]cysteine to the enzyme (Isd11p-independent step), although no persulfide was detected. Only after washing, lysis of mitochondria and Isd11p addition to the lysate, was the Nfs1p persulfide generated from prebound substrate (Isd11p-dependent step) (Figure 3A). Thus Nfs1p by itself was capable of binding the substrate cysteine in mitochondria but was unable to form persulfide until Isd11p was made available to the enzyme. The Isd11p-independent substrate binding and Isd11p-dependent persulfide formation on Nfs1p were also confirmed by experiments with purified proteins. Specifically, purified Nfs1p with prebound substrate [35S]cysteine was able to form the covalent persulfide only after the addition of purified Isd11p (Figure 3B). A possible mechanistic explanation of these findings is as follows.

The enzyme cofactor PLP bound at a conserved lysine residue of the enzyme is always available for interaction with the substrate cysteine, and this binding interaction is able to occur independently of Isd11p (Figure 9). However, in the absence of Isd11p, the active site cysteine residue on a protein loop at some distance away from the substrate-binding site was not close enough to initiate nucleophilic attack and persulfide formation. Isd11p binding to Nfs1p induces a conformational change in the enzyme that brings the active site cysteine residue and the substrate cysteine into proximity, thereby allowing persulfide formation at the active site cysteine residue. As part of the conformational change engendered by Isd11p binding, the Nfs1p active site cysteine residue becomes more resistant to alkylating agents. Thus Nfs1p alone in Isd11p▼ mitochondria was inactivated by 1 mM NEM (Figure 4) whereas the Nfs1p·Isd11p complex in WT mitochondria was more resistant, requiring a higher concentration (4 mM) of NEM for complete inactivation [26]. Likewise, purified Nfs1p by itself was also more sensitive to inactivation by NEM than the purified Nfs1p·Isd11p complex (Figure 5). Practically no persulfide formation was detected when purified Nfs1p was treated with 1 mM NEM and then supplemented with Isd11p. By contrast, persulfide formation by the preformed and purified Nfs1p·Isd11p complex was only slightly affected by 1 mM NEM and substantial inactivation occurred only after pretreatment with 4 mM NEM.

Our experimental system for studying Fe–S cluster synthesis in mitochondria involves addition of [35S]cysteine to isolated and metabolically active mitochondria. As described in the present study, the radioactive label can be detected on Nfs1p as a genuine persulfide in an Isd11p-dependent manner. Depending on conditions, the 35S label can also be identified on the scaffold protein as an Fe–S cluster intermediate [19], and on endogenous aconitase (Figure 8) or newly imported ferredoxin as an assembled Fe–S cluster [1921]. These novel assays together may now allow us to study step-by-step sulfur trafficking in mitochondria and how Isd11p might control the process. For example, imported Isd11p restored persulfide formation on Nfs1p in Isd11p▼ mitochondria. Surprisingly, at least three other proteins were also found to be radiolabelled after Isd11p import into these mitochondria (Figure 1B, compare lanes 2 and 3). Two of these proteins with estimated molecular masses of 24 and 22 kDa (indicated by arrowheads) were also detected in Nfs1p▼ mitochondria as background bands (Figure 1B, lane 6). By contrast, radiolabelling of a 12 kDa protein (indicated by an asterisk) required the presence of both Isd11p and Nfs1p. The 12 kDa* protein was radiolabelled in Isd11p▼ mitochondria only after Isd11p import (Figure 1B, compare lanes 2 and 3) or after addition of Isd11p to the mitochondrial lysate (Figure 3A, lanes 4–6). The protein was efficiently radiolabelled in WT mitochondria, but no significant radiolabelling was observed in Nfs1p▼ mitochondria (Figure 1B, compare lanes 1 and 6, and Figure 3A, compare lanes 1 and 7). The radiolabelled 12 kDa* protein is unlikely to be Isd11p, because Isd11p cannot form a covalent persulfide as it does not contain any cysteine residues.

It is tempting to speculate that the 12 kDa* protein serves as an intermediate recipient of persulfide from Nfs1p for downstream processes such as Fe–S cluster synthesis or tRNA thiolation. Such a notion is consistent with the effects of NEM pretreatment of mitochondria on Nfs1p-bound persulfide formation compared with radiolabelling of the 12 kDa* protein. When Isd11p▼ mitochondria were pretreated with NEM (1 mM) followed by neutralization with DTT, imported Isd11p failed to restore persulfide formation on Nfs1p and also radiolabelling of the 12 kDa* protein (Figure 4, lane 7). In the absence of Isd11p, the active site cysteine residue of Nfs1p was more susceptible to inactivation by NEM, thereby explaining the lack of persulfide formation on the enzyme. The absence of the radiolabelled 12 kDa* protein in these mitochondria appears to be due to Nfs1p inactivation and not due to NEM modification of the 12 kDa* protein itself. This view is supported by the observation that NEM (1 mM) pretreatment of WT mitochondria had only marginal effects on both the Nfs1p-associated persulfide and radiolabelling of the 12 kDa* protein (Figure 4, compare lanes 1 and 2). Taken together, the 12 kDa* protein could be an important recipient of the Nfs1p persulfide. The identity of the 12 kDa* protein remains to be determined.

In the cell, a certain flux of iron, cysteine, sulfur and Fe–S cluster intermediates must be maintained in order to supply newly synthesized apoproteins with sufficient cofactors and to repair damaged clusters. Both iron and sulfur are toxic when present in excess or unassembled and thus they must be tightly regulated to maintain the health and viability of cells. Several potential regulatory control points have now come to light. The Isd11p association with Nfs1p is necessary for activation of the cysteine desulfurase, and this might play a role in vivo in controlling the enzyme activity and the flow of sulfur into the Fe–S cluster assembly pathway. The formation of a larger complex consisting of human Nfs1–Isd11–Isu scaffold and frataxin has also been identified as a further activator of the cysteine desulfurase activity [40,41]. The delivery or release of the persulfide must also be regulated. The availability of ATP, GTP and NAD(P)H via effects on specific target proteins will probably contribute to regulating persulfide utilization. Correct integration of all of these regulatory inputs will be required to maintain healthy mitochondria and to avoid toxic damage.

Abbreviations

     
  • DTT

    dithiothreitol

  •  
  • DPD

    N,N-dimethyl-p-phenylenediamine dihydrochloride

  •  
  • IPTG

    isopropyl β-D-thiogalactopyranoside

  •  
  • Isd11p▼

    Isd11p-depleted

  •  
  • Isd11p

    mut, the tripeptide motif L15YK17 of Isd11p mutated to A15AA17

  •  
  • LB

    Luria–Bertani

  •  
  • NEM

    N-ethylmaleimide

  •  
  • Ni-NTA

    Ni2+-nitrilotriacetate

  •  
  • Nfs1p▼

    Nfs1p-depleted

  •  
  • ORF

    open reading frame

  •  
  • PLP

    pyridoxal phosphate

  •  
  • TCA

    trichloroacetic acid

  •  
  • WT

    wild-type

AUTHOR CONTRIBUTION

Alok Pandey performed persulfide formation, Fe–S cluster loading and import assays. Ramesh Golla isolated mitochondria from yeast strains with regulated expression of proteins and measured various enzyme activities. Heeyong Yoon did cellular and mitochondrial iron uptake measurements and colorimetric assays for cysteine desulfurase activity. Andrew Dancis and Debkumar Pain designed experiments and wrote the paper.

We thank Boominathan Amutha, Yajuan Gu and Elise R. Lyver for their valuable contributions in the early stage of this work.

FUNDING

This work was supported by the National Institute of General Medical Sciences [grant number GM087965 (to D.P. and A.D.)], the National Institute on Aging [grant number AG030504 (to D.P.)] and the National Institutes of Health [grant number R37DK053953 (to A.D)]. This work was also supported by the American Heart Association [grant number 09GRNT2260364 (to D.P.)].

References

References
1
Kiley
P. J.
Beinert
H.
The role of Fe-S proteins in sensing and regulation in bacteria
Curr. Opin. Microbiol.
2003
, vol. 
6
 (pg. 
181
-
185
)
2
Imlay
J. A.
Iron-sulphur clusters and the problem with oxygen
Mol. Microbiol.
2006
, vol. 
59
 (pg. 
1073
-
1082
)
3
Rouault
T. A.
The role of iron regulatory proteins in mammalian iron homeostasis and disease
Nat. Chem. Biol.
2006
, vol. 
2
 (pg. 
406
-
414
)
4
Stemmler
T. L.
Lesuisse
E.
Pain
D.
Dancis
A.
Frataxin and mitochondrial Fe-S cluster biogenesis
J. Biol. Chem.
2010
, vol. 
285
 (pg. 
26737
-
26743
)
5
Lill
R.
Function and biogenesis of iron-sulphur proteins
Nature
2009
, vol. 
460
 (pg. 
831
-
838
)
6
Kessler
D.
Enzymatic activation of sulfur for incorporation into biomolecules in prokaryotes
FEMS Microbiol. Rev.
2006
, vol. 
30
 (pg. 
825
-
840
)
7
Mueller
E. G.
Trafficking in persulfides: delivering sulfur in biosynthetic pathways
Nat. Chem. Biol.
2006
, vol. 
2
 (pg. 
185
-
194
)
8
Johnson
D. C.
Dean
D. R.
Smith
A. D.
Johnson
M. K.
Structure, function, and formation of biological iron-sulfur clusters
Annu. Rev. Biochem.
2005
, vol. 
74
 (pg. 
247
-
281
)
9
Cupp-Vickery
J. R.
Urbina
H.
Vickery
L. E.
Crystal structure of IscS, a cysteine desulfurase from Escherichia coli
J. Mol. Biol.
2003
, vol. 
330
 (pg. 
1049
-
1059
)
10
Kaiser
J. T.
Clausen
T.
Bourenkow
G. P.
Bartunik
H.-D.
Steinbacher
S.
Huber
R.
Crystal structure of a NifS-like protein from Thermotoga maritima: implications for iron sulphur cluster assembly
J. Mol. Biol.
2000
, vol. 
297
 (pg. 
451
-
464
)
11
Shi
R.
Proteau
A.
Villarroya
M.
Moukadiri
I.
Zhang
L.
Trempe
J. F.
Matte
A.
Armengod
M. E.
Cygler
M.
Structural basis for Fe-S cluster assembly and tRNA thiolation mediated by IscS protein-protein interactions
PLoS Biol.
2010
, vol. 
8
 pg. 
e1000354
 
12
Zheng
L.
White
R. H.
Cash
V. L.
Jack
R. F.
Dean
D. R.
Cysteine desulfurase activity indicates a role for NIFS in metallocluster biosynthesis
Proc. Natl. Acad. Sci. U.S.A.
1993
, vol. 
90
 (pg. 
2754
-
2758
)
13
Zheng
L.
White
R. H.
Cash
V. L.
Dean
D. R.
Mechanism for the desulfurization of L-cysteine catalyzed by the nifS gene product
Biochemistry
1994
, vol. 
33
 (pg. 
4714
-
4720
)
14
Rouault
T. A.
Tong
W. H.
Iron-sulfur cluster biogenesis and human disease
Trends Genet.
2008
, vol. 
24
 (pg. 
398
-
407
)
15
Li
J.
Kogan
M.
Knight
S. A. B.
Pain
D.
Dancis
A.
Yeast mitochondrial protein, Nfs1p, coordinately regulates iron-sulfur cluster proteins, cellular iron uptake, and iron distribution
J. Biol. Chem.
1999
, vol. 
274
 (pg. 
33025
-
33034
)
16
Kolman
C.
Söll
D.
SPL1-1, a Saccharomyces cerevisiae mutation affecting tRNA splicing
J. Bacteriol.
1993
, vol. 
175
 (pg. 
1433
-
1442
)
17
Land
T.
Rouault
T. A.
Targeting of a human iron-sulfur cluster assembly enzyme, nifs, to different subcellular compartments is regulated through alternative AUG utilization
Mol. Cell
1998
, vol. 
2
 (pg. 
807
-
815
)
18
Naamati
A.
Regev-Rudzki
N.
Galperin
S.
Lill
R.
Pines
O.
Dual targeting of Nfs1 and discovery of its novel processing enzyme, Icp55
J. Biol. Chem.
2009
, vol. 
284
 (pg. 
30200
-
30208
)
19
Amutha
B.
Gordon
D. M.
Gu
Y.
Lyver
E. R.
Dancis
A.
Pain
D.
GTP is required for iron-sulfur cluster biogenesis in mitochondria
J. Biol. Chem.
2008
, vol. 
283
 (pg. 
1362
-
1371
)
20
Amutha
B.
Gordon
D. M.
Dancis
A.
Pain
D.
Nucleotide-dependent iron-sulfur cluster biogenesis of endogenous and imported apoproteins in isolated intact mitochondria
Methods Enzymol.
2009
, vol. 
456
 (pg. 
247
-
266
)
21
Pain
J.
Balamurali
M. M.
Dancis
A.
Pain
D.
Mitochondrial NADH kinase, Pos5p, is required for efficient iron-sulfur cluster biogenesis in Saccharomyces cerevisiae
J. Biol. Chem.
2010
, vol. 
285
 (pg. 
39409
-
39424
)
22
Yoon
H.
Zhang
Y.
Pain
J.
Lyver
E. R.
Lesuisse
E.
Pain
D.
Dancis
A.
Rim2, a pyrimidine nucleotide exchanger, is needed for iron utilization in mitochondria
Biochem. J.
2011
, vol. 
440
 (pg. 
137
-
146
)
23
Wiedemann
N.
Urzica
E.
Guiard
B.
Müller
H.
Lohaus
C.
Meyer
H. E.
Ryan
M. T.
Meisinger
C.
Mühlenhoff
U.
Lill
R.
Pfanner
N.
Essential role of Isd11 in mitochondrial iron-sulfur cluster synthesis on Isu scaffold proteins
EMBO J.
2006
, vol. 
25
 (pg. 
184
-
195
)
24
Adam
A. C.
Bornhövd
C.
Prokisch
H.
Neupert
W.
Hell
K.
The Nfs1 interacting protein Isd11 has an essential role in Fe/S cluster biogenesis in mitochondria
EMBO J.
2006
, vol. 
25
 (pg. 
174
-
183
)
25
Shi
Y.
Ghosh
M. C.
Tong
W.-H.
Rouault
T. A.
Human ISD11 is essential for both iron-sulfur cluster assembly and maintenance of normal cellular iron homeostasis
Hum. Mol. Genet.
2009
, vol. 
18
 (pg. 
3014
-
3025
)
26
Pandey
A.
Yoon
H.
Lyver
E. R.
Dancis
A.
Pain
D.
Identification of a Nfs1p-bound persulfide intermediate in Fe-S cluster synthesis by intact mitochondria
Mitochondrion
2012
, vol. 
12
 (pg. 
539
-
549
)
27
Longtine
M. S.
McKenzie
III, A.
Demarini
D. J.
Shah
N. G.
Wach
A.
Brachat
A.
Philippsen
P.
Pringle
J. R.
Additional modules for versatile and economical PCR-based gene deletion and modification in Saccharomyces cerevisiae
Yeast
1998
, vol. 
14
 (pg. 
953
-
961
)
28
Sepuri
N. B. V.
Gordon
D. M.
Pain
D.
A GTP-dependent “push” is generally required for efficient protein translocation across the mitochondrial inner membrane into the matrix
J. Biol. Chem.
1998
, vol. 
273
 (pg. 
20941
-
20950
)
29
Gu
Y.
Gordon
D. M.
Amutha
B.
Pain
D.
A GTP:AMP phosphotransferase, Adk2p, in Saccharomyces cerevisiae: role of the C-terminus in protein folding/stabilization, thermal tolerance, and enzymatic activity
J. Biol. Chem.
2005
, vol. 
280
 (pg. 
18604
-
18609
)
30
Siegel
L. M.
A direct microdetermination for sulfide
Anal. Biochem.
1965
, vol. 
11
 (pg. 
126
-
132
)
31
Kennedy
M. C.
Beinert
H.
The state of cluster SH and S2− of aconitase during cluster interconversions and removal
J. Biol. Chem.
1988
, vol. 
263
 (pg. 
8194
-
8198
)
32
Munujos
P.
Coll-Canti
J.
Gonzalez-Sastre
F.
Gella
F. J.
Assay of succinate dehydrogenase activity by a colorimetric-continuous method using iodonitrotetrazolium chloride as electron acceptor
Anal. Biochem.
1993
, vol. 
212
 (pg. 
506
-
509
)
33
Siegel
L.
Englard
S.
Beef heart malic dehydrogenases. III. Comparative studies of some properties of M-malic dehydrogenase and S-malic dehydrogenase
Biochim. Biophys. Acta
1962
, vol. 
64
 (pg. 
101
-
110
)
34
Chouchani
E. T.
James
A. M.
Fearnley
I. M.
Lilley
K. S.
Murphy
M. P.
Proteomic approaches to the characterization of protein thiol modification
Curr. Opin. Chem. Biol.
2011
, vol. 
15
 (pg. 
120
-
128
)
35
Hochgräfe
F.
Mostertz
J.
Pöther
D.-C.
Becher
D.
Helmann
J. D.
Hecker
M.
S-Cysteinylation is a general mechanism for thiol protection of Bacillus subtilis proteins after oxidative stress
J. Biol. Chem.
2007
, vol. 
282
 (pg. 
25981
-
25985
)
36
Mustafa
A. K.
Gadalla
M. M.
Sen
N.
Kim
S.
Mu
W.
Gazi
S. K.
Barrow
R. K.
Yang
G.
Wang
R.
Snyder
S. H.
H2S signals through protein S-sulfhydration
Sci. Signaling
2009
, vol. 
2
 pg. 
ra72
 
37
Ghezzi
D.
Goffrini
P.
Uziel
G.
Horvath
R.
Klopstock
T.
Lochmüller
H.
D’Adamo
P.
Gasparini
P.
Strom
T. M.
Prokisch
H.
, et al. 
SDHAF1, encoding a LYR complex-II specific assembly factor, is mutated in SDH-defective infantile leukoencephalopathy
Nat. Genet.
2009
, vol. 
6
 (pg. 
654
-
656
)
38
Atkinson
A.
Smith
P.
Fox
J. L.
Cui
T.-Z.
Khalimonchuk
O.
Winge
D. R.
The LYR protein Mzm1 functions in the insertion of the Rieske Fe/S protein in yeast mitochondria
Mol. Cell. Biol.
2011
, vol. 
31
 (pg. 
3988
-
3996
)
39
Mihara
H.
Esaki
N.
Bacterial cysteine desulfurases: their function and mechanisms
Appl. Microbiol. Biotechnol.
2002
, vol. 
60
 (pg. 
12
-
23
)
40
Tsai
C.-L.
Barondeau
D. P.
Human frataxin is an allosteric switch that activates the Fe-S cluster biosynthetic complex
Biochemistry
2010
, vol. 
49
 (pg. 
9132
-
9139
)
41
Schmucker
S.
Martelli
A.
Colin
F.
Page
A.
Wattenhofer-Donzé
M.
Reutenauer
L.
Puccio
H.
Mammalian frataxin: an essential function for cellular viability through an interaction with a preformed ISCU/NFS1/ISD11 iron-sulfur assembly complex
PLoS ONE
2011
, vol. 
6
 pg. 
e16199
 

Supplementary data