Multiple Turnover Transfer of [2Fe2S] Clusters by the Iron-Sulfur Cluster Assembly Scaffold Proteins IscU and IscA*

IscU/Isu and IscA/Isa (and related NifU and SufA proteins) have been proposed to serve as molecular scaffolds for preassembly of [FeS] clusters to be used in the biogenesis of iron-sulfur proteins. In vitro studies demonstrating transfer of preformed scaffold-[FeS] complexes to apoprotein acceptors have provided experimental support for this hypothesis, but investigations to date have yielded only single-cluster transfer events. We describe an in vitro assay system that allows for real-time monitoring of [FeS] cluster formation using circular dichroism spectroscopy and use this to investigate de novo [FeS] cluster formation and transfer from Escherichia coli IscU and IscA to apo-ferredoxin. Both IscU and IscA were found to be capable of multiple cycles of [2Fe2S] cluster formation and transfer suggesting that these scaffold proteins are capable of acting “catalytically.” Kinetic studies further showed that cluster transfer exhibits Michaelis-Menten behavior indicative of complex formation of holo-IscU and holo-IscA with apoferredoxin and consistent with a direct [FeS] cluster transfer mechanism. Analysis of the dependence of the rate of cluster transfer, however, revealed enhanced efficiency at low ratios of scaffold to acceptor protein suggesting participation of a transient, labile scaffold-[FeS] species in the transfer process.

The biosynthesis of iron-sulfur proteins is a multistep process involving a number of specialized proteins that mediate [FeS] cluster formation and delivery to acceptor proteins (reviewed in Refs. [1][2][3]. One of the key features of the current view of the pathway is the proposed participation of protein scaffolds that function as the initial sites for formation of transient [FeS] clusters; these molecular scaffolds may serve to guide cluster assembly, protect nascent clusters in the cellular environment, and/or assist in cluster transfer to acceptor proteins. The concept of scaffold-mediated metallocluster formation emerged from studies on the nitrogenase MoFe protein (4 -6), and initial formation of [FeS] clusters on scaffold proteins was first pro-posed for NifU, a protein required for the biogenesis of [FeS] clusters in the nitrogenase Fe-protein (7). Subsequent studies of [FeS] cluster formation on other proteins implicated in ironsulfur protein biogenesis, the NifU-related protein IscU/Isu (8,9), IscA/Isa (10,11), and SufA (12), led to the suggestion that these proteins may also function as specific scaffolds for preassembly of [FeS] clusters.
In vitro studies demonstrating transfer of preformed scaffold- [FeS] complexes from NifU (13), IscU/Isu (9,14,15), IscA/ Isa (16 -18), and SufA (12,19) to several apoprotein acceptors have provided experimental support for their proposed roles as transient scaffolds. The mechanism of cluster transfer, however, is poorly understood. The structures of the scaffold- [FeS] complexes are not known, and the exact forms of [FeS] cluster(s) transferred have not been established. The solution NMR (20) and crystal structures (Protein Data Bank code 1SU0) of apo-IscU and the crystal structures of apo-IscA (21,22) show that cysteine residues presumed to be involved in binding to iron atoms of the clusters are located at the surface of the apo-forms of the scaffold proteins. The scaffold-bound [FeS] clusters are therefore likely to be at least partially exposed to solvent and accessible for transfer to acceptor proteins. Little is known about how the clusters are released, but there is some evidence for a direct transfer involving scaffold-acceptor complexes. Studies employing a modified form of human Isu containing a 57 Fe-labeled cluster showed that the label did not exchange with ferrous sulfate present in solution during cluster transfer to apo-ferredoxin (15), and studies with Escherichia coli IscA and SufA showed that the iron chelator bathophenanthroline sulfonate does not interfere with cluster transfer to apo-BioB (12). These findings are consistent with a mechanism in which transfer occurs without a significant degree of cluster disassembly and reassembly and suggest that clusters may be captured immediately upon release from the scaffold protein or may be transferred directly in a scaffoldacceptor protein complex. Evidence that complex formation can occur has been obtained for IscA-ferredoxin and SufA-BioB pairs using affinity chromatography (16,19) and for Isu-and Isa-ferredoxin pairs using chemical cross-linking (9,11). However, the nature of the scaffold-acceptor interactions has not been characterized, and it has not been firmly established that complex formation is required for cluster transfer.
Although these in vitro studies provide support for scaffold proteins in [FeS] cluster assembly and delivery, the cluster transfers described to date have been relatively inefficient. In some cases the overall yield of cluster transfer has approached 100%, but the transfer rates observed have generally been slow (k obs Ͻ0.1 min Ϫ1 ) and in some cases much slower than the cell division time of the source organism. This raises the question of whether the observed transfers reflect the physiological mechanism of cluster formation and transfer or whether they are simply a thermodynamic consequence of the greater stability of * This work was supported by the Italian Ministry of University and Research, Progretti di rilevanza nazionale Grant Prot. 2004058243 (to F. B.) and National Institutes of Health Grant GM54264 (to L. E. V.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. the acceptor protein-[FeS] complexes compared with that of the scaffold- [FeS] complexes. In addition, none of the studies reported has demonstrated that the scaffold protein is able to function in the manner expected for a carrier-type catalyst, i.e. to mediate multiple cluster transfer reactions. In each study a stoichiometric or excess amount of preformed scaffold- [FeS] complex was used to deliver cluster to the acceptor protein, and turnover or recycling of the scaffold proteins has not been demonstrated. This raises the question of whether [FeS] complexes of scaffold proteins can function sufficiently rapidly to support the iron-sulfur protein requirements of the cell.
In the studies described herein we have developed an in vitro assay system that allows for real-time monitoring of [FeS] cluster formation in solutions containing both scaffold and acceptor proteins. We have used this system to investigate de novo [FeS] cluster formation and transfer from the E. coli proteins IscU and IscA to E. coli apo-ferredoxin. This system has allowed us to observe multiple [FeS] cluster transfer cycles and to investigate the dependence of transfer rate on scaffold and acceptor protein concentrations.

MATERIALS AND METHODS
Protein-Recombinant forms of E. coli apo-IscU (23), apo-IscA (22), and ferredoxin (24) were prepared according to previously published procedures, and protein concentrations were determined as described therein. Concentrations of the holo-forms of IscU, IscA, and ferredoxin are reported in terms of the [2Fe2S] complex of each protein. IscU behaves as a dimer in both the apo-and holo-forms (8,23), and concentrations are therefore reported in terms of the (IscU) 2 -[2Fe2S] complex. The oligomeric state of IscA is less well defined (11,16,18), but the crystal structure of apo-IscA suggests that two [2Fe2S] clusters can be bound to a tetramer (22), and concentrations are therefore reported in terms of an (IscA) 2 -[2Fe2S] complex. Holo-ferredoxin contains a single [2Fe2S] cluster (24,25).
Sample Handling and Cluster Reconstitution-Sample manipulations, including reagent preparation and chromatography, were carried out anaerobically under argon in septum-capped vials, cuvettes, or columns, and stainless steel needles, tubing, and cannulae were used for sample transfers. Unless otherwise specified the buffer used was 0.1 M Tris-HCl, pH 8.0, containing 5 mM dithiothreitol (TD buffer), and the temperature was maintained at 25°C.
Holo-IscU was prepared by three sequential additions at 5-min intervals of 0.5 eq of ferric ammonium citrate and lithium sulfide to apo-IscU; the final solution thus contained 1.5-fold greater concentrations of iron and sulfide than required for formation of the (IscU) 2 -[2Fe2S] complex. Holo-IscA was prepared similarly using ferrous ammonium sulfate in place of the ferric salt. For both proteins formation of [2Fe2S] clusters was monitored by visible region CD spectroscopy (cf. Fig. 1). In some cases (IscU) 2 -[2Fe2S] and (IscA) 2 -[2Fe2S] were purified by sequential gel permeation and ion exchange anaerobic chromatography on small G25 and DE52 columns; both holo-forms were found to be stable for at least 24 h under anaerobic conditions at 0°C. Elemental analyses of holo-IscU and holo-IscA for iron and sulfide were consistent with the presence of one [2Fe2S] cluster per dimer.
Ferredoxin was isolated as the [2Fe2S] complex (24). Apo-ferredoxin was prepared by precipitating the holo-form in 10% trichloroacetic acid containing 10 mM dithiothreitol for 10 min at 0°C. The sample was collected by centrifugation, washed twice in water under anaerobic conditions, and dissolved anaerobically in TD buffer.
Cluster Transfer Studies-Unless otherwise specified, apo-IscU (or apo-IscA) and apo-ferredoxin in TD buffer were mixed at 25°C in a 1-ml cuvette with 0.2 mM ferric (or ferrous) salt, typically equivalent to five times the amount of apo-ferredoxin (i.e. 2.5-fold that required for formation of the [2Fe2S]-ferredoxin). Lithium sulfide was used as the source of sulfide rather than the physiological donor cysteine and IscS (8) to ensure that sulfide availability was not rate-limiting. Reactions were initiated by the addition of five equivalents of lithium sulfide (0.2 mM final concentration), and the kinetics of cluster assembly and transfer were monitored by CD spectroscopy. Control experiments, in the absence of scaffold protein, were also carried out to determine the background rates of ferredoxin-[2Fe2S] formation.
Analytical Method-Analyses for iron and sulfide were carried out as described previously (26). CD measurements were recorded at 25°C in 1-cm path anaerobic cuvettes using a Jasco J-810 spectropolarimeter.

Circular Dichroism of [FeS] Complexes-Previous
[FeS] cluster transfer studies utilized UV-visible absorption spectroscopy, analytical gel electrophoresis, or biochemical assays to determine the amount of acceptor complex formed. The electrophoretic and biochemical methods employed require interrupting the transfer reaction and/or separation of scaffold and acceptor proteins and do not allow for real-time characterization of reaction progress (9, 13-15, 17, 18). UV-visible spectroscopy has been used to directly monitor cluster transfer with IscA and SufA (12,16,19), but the general similarity and the absence of protein-specific signature features of the absorption spectra of [FeS] complexes makes it difficult to determine whether intermediate species having different coordination and/or geometry might occur during the transfer reaction.
CD spectra of metal complexes are generally more complex than absorption spectra and provide increased sensitivity for detecting differences in cluster properties. The near UV-visible region CD of the [2Fe2S] complex of E. coli IscU (39) is qualitatively similar to that of E. coli ferredoxin-[2Fe2S] (24), consistent with similar cluster coordination, but the CD intensity of IscU 2 -[2Fe2S] is weaker suggesting different cluster environments in the two proteins. The CD of IscA has not been reported, but models of [2Fe2S] complexes of IscA based on the crystal structure (22) suggest that the cluster environment is likely to be much different from that of ferredoxin (25). We sought to determine whether CD could provide a more sensitive means of mon- itoring cluster transfer from IscU and/or IscA to ferredoxin. Fig. 1 shows near UV-visible region CD spectra of purified [2Fe2S] forms of IscU, IscA, native ferredoxin, and ferredoxin reconstituted using the scaffold proteins. Native ferredoxin-[2Fe2S] was isolated as described previously (24) complexes. An excess of apo-ferredoxin (1.6 eq) was used to favor complete cluster transfer from the scaffold proteins. To determine whether IscU and IscA are able to mediate multiple cluster transfer cycles requires: 1) use of substoichiometric amounts of the scaffold protein relative to apo-ferredoxin, 2) a system for continuous regeneration of the scaffold-[2Fe2S] complex, and 3) that the rate of cluster assembly on the scaffold is greater than the rate of unassisted ferredoxin cluster assembly. As described under "Materials and Methods" we found that IscU 2 -[2Fe2S] was formed in good yield using ferric ammonium citrate and lithium sulfide in TD buffer, and we monitored the rate of cluster formation at 25°C by singlewavelength CD measurements at 540 nm. Supplemental Fig.  1 shows that using 100 M apo-IscU, formation of IscU 2 -[2Fe2S] is Ͼ90% complete within 5 min with k obs Х 0.3 min Ϫ1 . Spontaneous formation of ferredoxin-[2Fe2S] under similar conditions, in contrast, is much slower and requires several hours to reach completion (see below). Therefore under these conditions formation of holo-IscU is not expected to be rate-limiting, and the rate of holo-ferredoxin formation in mixtures of apo-IscU and apo-ferredoxin will reflect the rate of cluster transfer from holo-IscU to apo-ferredoxin and/or ferredoxin cluster maturation.
Initial studies to test for a catalytic role of IscU in mediating multiple cluster transfers were carried out using apo-IscU 2 and apo-ferredoxin in a molar ratio of 0.1 to 1. In the experiment shown in Fig. 2 apo-ferredoxin alone or apo-ferredoxin and apo-IscU were pre-equilibrated in the presence of ferric ammonium citrate, and cluster formation was initiated by addition of lithium sulfide. IscU was found to significantly enhance cluster assembly on ferredoxin. The initial rate of ferredoxin-[2Fe2S] formation is Х9-fold greater in the presence of IscU than in its absence suggesting that scaffold-mediated preassembly and transfer is more efficient than direct assembly on the ferredoxin. The final yield of holo-ferredoxin is also greater (Х100 versus Х75%) suggesting that the scaffold-mediated process may prevent unproductive side reactions from occurring. The CD spectra at different stages of completion of the reaction are qualitatively similar and exhibit distinct isodichroic points at and apoferredoxin with rapid association/dissociation kinetics followed by slower transfer of the [FeS] cluster. The apparent K m for the reaction is Х27 M suggesting a relatively weak interaction between the two proteins. The maximal turnover number extrapolated to saturating levels of apo-ferredoxin is Х0.21 mol holo-ferredoxin/mol IscU 2 min Ϫ1 indicating that cluster transfer in this system is a relatively slow process.
Catalysis of Cluster Transfer by IscA-Studies were also carried out to determine whether IscA is able to catalyze multiple cluster transfer cycles. Because cluster assembly on IscA is more efficient with ferrous salts, apo-IscA and apo-ferredoxin were equilibrated with ferrous ammonium sulfate, and the reaction was initiated by addition of lithium sulfide. Fig. 4 shows a time course experiment employing a molar ratio of apo-IscA 2 :apo-ferredoxin of 0.1 to 1. The stimulation of the initial rate of ferredoxin-[2Fe2S] formation (Х4-fold) is not as great as that observed with IscU, but enhancement continues throughout the reaction consistent with IscA mediating multiple catalytic cycles. The final yield (18 h) of holo-ferredoxin in the presence of IscA is also greater than in its absence (Х80 versus Х65%) suggesting that the scaffold-mediated process is somewhat more effective. As found for IscU, the CD spectra at In addition, the maximal extrapolated turnover number (Х0.03 min Ϫ1 ) is ϳ7-fold slower than that observed with IscU 2 -[2Fe2S] (Х0.2 min Ϫ1 ) indicating that IscA is a less effective catalyst for cluster transfer to apo-ferredoxin. The lower affinity and poorer catalysis are consistent with the finding that the final yield obtained with IscA (Х75%, Fig. 4) is less than that obtained with IscU (Х100%, Fig. 2).  Table I. Daisy wheel symbols (circles for ISU1 and square for ISA1) were used to distinguish these data (Refs. 15 and 17) from our data (this study). Inset, double logarithmic plot for the dependence of the rate of holo-ferredoxin formation on the molar ratio between the apo-IscU 2 or apo-IscA 2 and apo-ferredoxin. Lines shown represent fits of each set of data to a linear regression function.
Additional studies with other apo-protein acceptors will be required to determine whether the reduced effectiveness of IscA compared with IscU reflects a fundamental difference in the mechanism of cluster transfer by IscA and IscU or results from differences in their specific apo-protein acceptor preferences. There is some indication that IscA 2 -[2Fe2S] may be more stable than IscU 2 -[2Fe2S], and this could contribute to differences in rates of cluster transfer. In vitro studies have shown that clusters on IscU 2 -[2Fe2S] can be transferred to apo-IscA, whereas IscA 2 -[2Fe2S] will not transfer its cluster to apo-IscU (19). However, the exact cellular role of IscA has not been established. In addition to forming [FeS] complexes, IscA binds iron and may play a role in iron delivery to IscU for cluster assembly rather than as an alternate [FeS]-scaffold protein (27,28).
Effect of Scaffold Concentration-Additional cluster transfer experiments were carried out using molar ratios of apo-IscU and apo-IscA to apo-ferredoxin ranging from 0.05:1 to 1.2:1. An analysis of the results in terms of the observed specific activity (mol holo-ferredoxin/mol scaffold/min) versus the ratio of IscU 2 or IscA 2 to apo-ferredoxin is presented in Fig. 6. Surprisingly, the specific activities observed with both IscA and IscU were found to vary with the scaffold:apo-ferredoxin ratio. Higher specific activities were observed at lower ratios, and the specific activity increased Ͼ10-fold as the ratio of scaffold:acceptor decreased from about 1:1 to less than 0.1:1. Reliable rate measurements at low IscA 2 :apo-ferredoxin ratios could not be obtained, but the higher activity of IscU made it possible to determine rates at IscU 2 :apo-ferredoxin ratios as low as 0.05:1. The increased activity of IscU at low ratios is especially evident at IscU 2 :apo-ferredoxin ratios lower than 0.3:1. Specific activities increased ϳ20-fold from Х0.01 to Х0.2 mol holo-ferredoxin/mol IscU 2 /min as the scaffold:acceptor ratio decreased from 1:1 to 0.05:1. The specific activities observed using these low amounts of apo-IscU are also greater than those reported for transfer of preformed clusters in other systems. Table I gives a comparison of the specific activities observed in our multiple turnover assays with specific activities calculated from rates of transfer from preformed Isu and Isa complexes reported by others. The specific activity using a low ratio of apo-IscU is 10 -30-fold higher than single-transfer reactions using preformed Isu-[FeS] complexes. IscA is less effective than IscU, and the specific activities we observed in multiple turnover reactions differed only slightly from the single-cluster transfer activities using preformed Isa-[FeS] clusters.
The basis for the observed enhanced cluster transfer at low scaffold-to-acceptor ratios is not known. One explanation is that cluster transfer may involve a labile scaffold-[FeS] species that is transiently formed during initial cluster assembly, and high levels of apo-protein acceptor may favor capture and transfer of this species prior to its maturation to a more stable scaffold-[2Fe2S] complex. A more labile species may also be favored at low scaffold concentrations.  (10), and intermediate [FeS] species may participate in the transfer process. Solution NMR studies of apo-IscU from Thermotoga maritima (29,30), human, 1 yeast 1 and E. coli 2 indicate presence of significant dynamic flexibility in the protein backbone, and a specific IscU 2 -[2Fe2S] species with a conformation that differs from the major component could be involved in interactions with acceptor proteins and in cluster transfer.
Cellular Mechanism of Cluster Transfer-The maximal rates of scaffold-mediated cluster transfers observed in the in vitro experiments reported to date are slow (Ͻ1 min Ϫ1 ), and participation of additional cellular components may be required to achieve efficient cluster transfer in vivo. Molecular chaperones have been implicated in iron-sulfur protein biosynthesis (cf. Refs. 1-3), and these could play a role in generating or stabilizing reactive scaffold-[FeS] species and thereby facilitate cluster transfer. Recently a report appeared describing studies on the effect of the heat shock chaperone DnaK from T. maritima on transfer of preformed [FeS] complexes of T. maritima IscU or human Isu to human ferredoxin (31). The results from that study suggested that DnaK has only a minor influence on scaffold-[FeS] stability or transfer, and nucleotides and the co-chaperone DnaJ did not elicit effects expected for a specific, chaperone-assisted system (31). Other bacteria and eukaryotes, however, have specialized chaperone/co-chaperone systems that display specific interactions with IscU/Isu homologs (32)(33)(34)(35)(36)(37)(38), and these may function to regulate [FeS] cluster transfer in vivo. complexes were used as cluster donors. The specific activity for cluster transfer using preformed human holo-Isu 2 and human apo-ferredoxin (Fdx) and using preformed S. pombe holo-Isu 2 and S. pombe apo-ferredoxin were calculated from Table 1 of Ref. 15. The specific activity for cluster transfer using preformed S. pombe holo-Isa 2 and S. pombe apo-ferredoxin was calculated from Table 1