From Iron and Cysteine to Iron-Sulfur Clusters: the Biogenesis Protein Machineries

MARC FONTECAVE,^1,2,3* BÉATRICE PY,^4,5 SANDRINE OLLAGNIER de CHOUDENS,^1,2,3 AND FRÉDÉRIC BARRAS^4,5

[SECTION EDITOR: T. BEGLEY]

Posted August 1, 2008

CNRS UMR 5249¹ and CEA, DSV, iRTSV, Laboratoire de Chimie et Biologie des Métaux, 17 Avenue des Martyrs,² F-38054 Grenoble; Université Joseph Fourier, F-38000 Grenoble³; Université de la Méditerranée, Aix-Marseille II, Campus de Luminy, 70 Rte Léon Lachamp, 13009 Marseille⁴ ; and Laboratoire de Chimie Bactérienne, UPR 9043, 31 Chemin Joseph Aiguier, 13402 Marseille Cedex 20⁵; France

*Corresponding author. Mailing address: CEA, DSV, iRTSV, Laboratoire de Chimie et Biologie des Métaux, 17 Avenue des Martyrs, F-38054 Grenoble, France. Phone: 0033438789103, Fax: 0033438789124, E-mail: This e-mail address is being protected from spambots. You need JavaScript enabled to view it

Iron-sulfur (Fe-S) clusters are among the most structurally and functionally versatile cofactors in biology. Their controlled insertion into protein active sites is a complex mechanism which requires dedicated protein machineries. This chapter describes the two main systems, namely, the Isc (iron-sulfur cluster) and Suf (sulfur assimilation) systems, utilized by Escherichia coli and Salmonella for the biosynthesis of Fe-S clusters, as well as other proteins presumably participating in this process.

INTRODUCTION

Fe-S Clusters

Protein-bound Fe-S clusters, which are polynuclear combinations of iron and sulfur atoms, are among the most structurally and functionally versatile cofactors in biology. The Fe-S clusters were initially discovered, in the mid-1960s, as catalytic centers in electron transfer enzymes participating in important bioenergetic functions such as photosynthesis and respiration. They are present in organisms that are found in the three domains of life, bacteria, archaea, and eukaryotes, wherein they participate in a variety of cellular processes, such as DNA replication, DNA repair, transcription regulation, RNA modification, metabolite biosynthesis, and bioenergetic processes. The 2Fe-2S, 3Fe-4S, and 4Fe-4S clusters depicted in Fig. 1 are the most common ones. In these complexes, Fe ions are bound to the polypeptide by heteroatoms of amino acid side chains (such as sulfur from cysteine) and are linked to one another through sulfide bridges, resulting in materials with unique spectroscopic and chemical properties.

Fig. 1The three most frequently occurring Fe-S clusters. The orange balls illustrate the sulfur atoms. Cys indicates the thiolates from the protein cysteine residues that coordinate the clusters.

The great diversity of the chemical properties of Fe-S clusters has been exploited by proteins for assistance in biological processes in different ways, listed hereafter (56). These clusters are ideal for catalyzing intra- and intermolecular electron transfer processes because they can access various redox states and the potentials of the corresponding redox couples can be finely tuned by the coordination environment, by the electronic properties of the protein site where they are anchored, and by hydrogen bonding (58). Redox potentials for Fe-S clusters vary from more than 500 mV to less than −500 mV, a range larger than those for any other simple redox cofactors. Those Fe-S clusters with very low redox potentials can be used for the reduction of redox-resistant substrates, and some proteins have exploited this property for redox catalysis. A fascinating example is illustrated by the enzymes of the so-called radical–S-adenosylmethionine (SAM) superfamily. In these enzymes, a 4Fe-4S cluster is used to inject one electron at a low potential into SAM and to generate reactive free radicals in a great variety of biosynthetic and metabolic reactions (19, 55, 106, 110, 116). The Fe-S clusters can also be designed to allow the access of small compounds to ferric sites endowed with strong Lewis acidity properties. Because this access results in strong and selective polarization and the activation of substrate bonds, such clusters can be used in non-redox catalysis, as illustrated by dehydratases, with aconitase as a prototype (11). Reversible interconversions between cluster forms with different redox states or with different nuclearities, as a consequence of a reaction with a substrate, represent a unique property of this class of biological metallosites which can be exploited, for instance, in the form of switches to control DNA and mRNA binding activity. As a matter of fact, a number of Fe-S proteins function as sensors of oxygen, superoxide, nitric oxide, or iron for the purpose of gene expression regulation and adaptation to changing living conditions (15, 25, 37, 66, 128).

Approximately 100 Fe-S proteins have been isolated from E. coli so far (33). Table 1 presents the 88 proteins which have been unambiguously shown, at least by iron and sulfur content analysis and/or spectroscopy (UV-visible light, electronic paramagnetic resonance, or Mössbauer), to contain a cluster, and Fig. 2 indicates the locations of the corresponding genes. Of the cluster types observed, the 2Fe-2S and 3Fe-4S proteins are remarkably minor, representing approximately 10%, whereas the 4Fe-4S proteins account for 90% (33). Furthermore, the electron transfer systems correspond to more than 60% of this list, whereas each one of the other functions accounts for a small fraction (nonredox enzymes and redox enzymes constitute about 10% each). Finally, when the E. coli genome is searched for characteristic sequence motifs, one can find an additional 100 putative Fe-S proteins (M. Fontecave and Y. Vandenbrouck, unpublished results). This means that the number of Fe-S proteins may be approximately 5% of the total number of proteins in E. coli, which is much larger than generally speculated.

Fig. 2E.coli genes known to encode Fe-S proteins and components involved in Fe-S biosynthesis. The schematic representation of the E.coli chromosome shows the relative locations of the genes encoding Fe-S proteins (outside) and Fe-S biosynthesis components (inside).

Table 1List of characterized Fe-S cluster proteins in E.coli

The unique redox properties of the clusters have a major drawback: they make most of the Fe-S enzymes sensitive to oxidative stress in the form of molecular oxygen, nitric oxide, and reduced oxygen species such as superoxide and hydrogen peroxide. Reaction with these compounds in general results in cluster damage and degradation, the loss of enzyme activity, and iron release, which subsequently exacerbates the oxidative stress by the Fenton reaction (52, 54). As a consequence, microbial cells have set up exquisite mechanisms, still ill-defined, to synthesize and protect these critical cofactors during aerobic life.

Fe-S Cluster Biosynthesis: Cysteine Desulfurases and Scaffold Proteins

As early as the 1970s, it was shown that Fe-S clusters can be synthesized in vitro during the reaction of an apoprotein with ferrous and sulfide salts, but it was not until the 1990s that it was discovered that cells employ complex protein machineries for cluster biosynthesis (53). Since then, there has been an impressive effort to understand the mechanism of Fe-S cluster assembly, mainly through extensive genetic and biochemical studies carried out with E. coli, Salmonella, and Saccharomyces cerevisiae.

The biosynthesis of Fe-S clusters is assumed to conform to the following rules. First, specific donor proteins, rather than free iron and free sulfur, are the source of cluster components under in vivo conditions. Accordingly, the different assembly machineries identified to date share the involvement of acysteine desulfurase which allows the utilization of l-cysteine as a source of sulfur atoms (Fig. 3) (88). This fascinating reaction implies a rupture of the terminal C-S bond in the l-cysteine. Cysteine desulfurases are pyridoxal phosphate (PLP)-dependent enzymes which degrade l-cysteine into l-alanine and sequester the released sulfur atom in the S⁰ redox state on a specific cysteine residue, in the form of a persulfide moiety (Fig. 3). The mechanism of this reaction has been extensively studied and characterized (88, 152, 153). A version of this mechanism is represented in Fig. 4, which emphasizes the crucial role of the PLP cofactor during catalysis (35). The first step, following the binding of the l-cysteine substrate to the active site of the protein, consists of the generation of an intermediate aldimine adduct by the reaction of the amino group of the substrate with PLP. Then the aldimine converts into a ketimine adduct. The second step involves an essential cysteine in the active site which, in the deprotonated cysteinate form, carries out a nucleophilic attack on the electrophilic sulfur atom of the ketimine intermediate. This process generates a stable protein-bound persulfide, together with a novel ketimine adduct. The hydrolysis of the latter results in the formation and release of l-alanine, one of the final products of the reaction, and regenerates PLP for a new catalytic cycle. The terminal S⁰ atom of the persulfide can jump from the desulfurase (the donor) onto another protein (the acceptor) by a transpersulfuration reaction, which implies a nucleophilic attack on the persulfide by a cysteinate of the acceptor (Fig. 5) (101, 123, 132, 133). This process provides an exquisite mechanism for sulfur atoms to travel within the cell from one protein to another over long distances, without release in solution. In the case of Fe-S cluster biosynthesis, it is assumed that the sulfur atoms from the cysteine desulfurase end up at cysteine residues of the scaffold protein, presumably waiting for iron atoms for cluster assembly (see below).

Fig. 3Cysteine desulfurase activity. Cysteine desulfurase is a PLP-dependent enzyme which binds a sulfur atom, provided by l-cysteine, in a persulfide form. CD, cysteine desulfurase. PLP, pyridoxal phosphate.

Fig. 4Detailed mechanism for enzymatic PLP-dependent cysteine desulfuration. PLP reacts with the l-cysteine substrate to generate an intermediate aldimine adduct which converts into a ketimine adduct. Then the essential catalytic cysteine of the cysteine desulfurase performs a nucleophilic attack on the PLP-cysteine substrate, leading to a protein-bound persulfide. Hydrolysis generates l-alanine and PLP. CD-S—H, cysteine desulfurase with its active cysteine residue. P-O stands for phosphate group.

Fig. 5Transpersulfuration reaction. The terminal sulfur atom of a protein-bound persulfide (donor) can be transferred to a cysteine of another protein (acceptor).

The source of Fe is a more obscure issue. Whether the low-molecular-weight iron pool or, instead, specialized iron binding proteins of still unidentified natures are involved in this process is unknown. A section below discusses the various potential iron donor proteins. For in vitro experiments, in general, ferrous salts are used during the assembly of Fe-S clusters, even though this approach is unlikely to reflect the physiological conditions.

Second, clusters are assembled primarily within a scaffold protein, from which they are transferred as a whole to recipient apoproteins (Fig. 6). Thus, scaffold proteins bind clusters only transiently. This cluster lability makes the clusters difficult to characterize biochemically. Furthermore, it explains why in general these proteins are isolated in an apoform and bind a mixture of 2Fe-2S and 4Fe-4S clusters during chemical reconstitution, with no clear indication whether this pattern reflects incomplete cluster assembly or some cluster degradation (1, 67, 102, 105). The mechanism of cluster assembly is still a matter of controversy and debate. The fact that sulfur atoms can be directly transferred from cysteine desulfurases to scaffold proteins supports a mechanism in which the latter bind sulfur atoms first and iron atoms afterwards (34, 35, 123, 132, 133, 141). However, in some cases, scaffold proteins have been shown to be able to bind ferrous or ferric iron in the absence of sulfur, suggesting the "iron first, sulfur second" mechanism (34, 100). It is important to note that, with only one exception (123), the assembly of clusters in such sequential protocols in vitro has failed. Cluster transfer from the scaffold protein to the apoprotein target is likely to be direct and, thus, involve an intermediate protein-protein complex, providing a mechanism for avoiding the release of iron and sulfur in solution (103). The molecular details of this reaction are largely unknown.

Fig. 6Fe-S cluster biogenesis. The assembly of a cluster within a scaffold protein and the transfer to a target apoprotein (apo-target) are depicted. Holo-target, holoform target.

In summary, the main functions of Fe-S cluster assembly protein machineries are (i) to mobilize Fe and S atoms from their storage sources, (ii) to assemble them into an Fe-S cluster form, and (iii) to transfer the assembled clusters to their final protein destinations (Fig. 6). All these reactions are tightly controlled so that Fe and S atoms are not liberated in solution during the process. It is clear that in a given cell there are many more Fe-S proteins to maturate than scaffold proteins, so the latter must be exquisitely designed to allow efficient cluster delivery to a large number of apoprotein targets, which differ in size, localization, and structure. How this is achieved is an intriguing question, considering the importance of protein-protein interactions during the process. On the other hand, it seems that in a few cases a specific cluster assembly system is required, as shown with the nitrogen fixation machinery in Azotobacter vinelandii (NIF), which specifically maturates nitrogenase (29, 53, 57, 153).

E. coli and Salmonella have two main assembly machineries, referred to as the Isc (iron sulfur cluster) and Suf (sulfur assimilation) systems, whose corresponding genes are organized in operons (7, 56). Isc-like systems are widespread in a number of eubacteria and archaea and are also found in the mitochondria of eukaryotes, whereas the Suf-like systems are present in prokaryotes, chloroplasts, and parasites. In addition, several Fe-S cluster biogenesis ancillary factors whose activities may be either redundant with or complementary to those of the Isc and Suf machineries have been identified. Such a large repertoire of cluster biogenesis-assisting factors is very likely to allow E. coli and Salmonella to maturate Fe-S proteins under a large scope of environmental conditions, some of them being highly detrimental to Fe-S clusters, such as, for instance, aerobiosis.

THE Isc SYSTEM OF E.COLI AND SALMONELLA

The isc genes involved in Fe-S cluster assembly were discovered in Azotobacter vinelandii in D. Dean’s laboratory in 1998 (151). The importance of the Isc system for Fe-S cluster biosynthesis in E. coli was demonstrated 1 year later by Y. Takahashi and collaborators (97, 135). It was indeed shown that the production of reporter heterologous holoferredoxins is dramatically increased by the coexpression of the isc gene cluster (97). Furthermore, by using plasmids in which each open reading frame was individually inactivated, it was shown that all of the isc genes except iscA are required under normal growth conditions for the production of ferredoxins of the holoform (135, 139). Then a number of studies, mainly with E. coli and a few with Salmonella, associated a series of phenotypes, including reduced growth rates and a requirement for nicotinic acid and thiamine, with mutations in isc genes (69, 120, 130). These phenotypic defects are caused in part by reduced activities of Fe-S enzymes, such as NadA (quinolinate synthase) or ThiH, involved in NAD and thiamine biosynthesis, respectively (20, 75, 86, 104). The Isc system comprises seven genes, organized as an operon in the following order: iscRSUA-hscBA-fdx. We discuss below the properties of each of the proteins of the Isc system. IscR, a transcriptional regulator which controls the expression of over 40 genes in the chromosome, is discussed in a separate section dedicated to genetic regulation.

IscS, the Cysteine Desulfurase

The deletion of the iscS gene in E. coli leads to a reduced growth rate in rich medium (28, 69, 120, 139). The iscS mutant has drastically reduced activities (2 to 10% of the wild-type levels) of several Fe-S enzymes such as aconitase, 6-phosphogluconate dehydratase, fumarase A, NADH dehydrogenase, and succinate dehydrogenase (28, 120). This finding strongly supports the concept that IscS is the major source of sulfur for Fe-S cluster biogenesis in E. coli under normal growth conditions. Moreover, the iscS mutant requires thiamine and nicotinic acid for growth in minimal medium under aerobiosis, while under anaerobiosis, nicotinic acid only is required (120). Two key enzymes of the thiamine biosynthetic pathway are potentially linked to IscS: ThiH, a radical-SAM enzyme, because it contains an Fe-S cluster whose assembly may depend on IscS (75, 86), and ThiS, whose terminal carboxylate is sulfurated by the cysteine desulfurase activity of IscS (69, 111). Nicotinic acid is used by the iscS mutant to compensate for the inactivation of NadA, the quinolinate synthase enzyme catalyzing one step of NAD biosynthesis, probably due to defects in the assembly of its catalytically essential 4Fe-4S cluster (20, 104). In addition to the auxotrophy for thiamine and nicotinic acid, the reduced growth rate of the iscS mutant is likely to be due to other effects since IscS is also involved in other metabolic pathways, among them the thiolation of tRNA, in both E. coli and Salmonella (65, 68, 98).

IscS from E. coli is a homodimer of 90 kDa which has been structurally characterized previously (24). The three-dimensional structure shows that the PLP cofactor is attached to the protein through an internal aldimine Schiff base with Lys206. The position of the conserved cysteine (Cys328), which binds sulfur atoms in a persulfide form, is intriguingly located 17 Å away from the PLP cofactor, suggesting that a significant conformational change must occur during catalysis in order for Cys328 to react with the PLP-bound cysteine substrate and generate the protein-bound persulfide (24) (Fig. 4). On the other hand, Cys328 is in a solvent-accessible mobile loop, thus facilitating the transfer of S atoms from persulfide to a cysteine of an acceptor protein. Cys328 has been shown previously, by site-directed mutagenesis, to be essential for cysteine desulfurase activity (89).

IscU, the Scaffold Protein

In E. coli, the inactivation of the iscU gene leads to a slow-growth phenotype and reduced activities of many Fe-S proteins to 10 to 60% of the wild-type levels (28, 139). Recombinant IscU from A. vinelandii is a homodimeric protein which, upon the addition of Fe²⁺ and S²⁻, provided by l-cysteine and IscS, sequentially binds one 2Fe-2S cluster per dimer, then one 2Fe-2S cluster per monomer, and finally, one 4Fe-4S cluster per dimer (Fig. 7) (1). The clusters in IscU have the properties of being both reductively labile and sensitive to oxygen. There are many observations that support the hypothesis that a direct interaction between IscU and IscS allows for a direct transfer of sulfane S⁰ atoms. First, sulfur atom transfer between E. coli IscS and IscU has been demonstrated previously using mass spectrometry and radioactively labeled cysteine (132, 141). Second, E. coli IscU induces a sixfold increase in the cysteine desulfurase activity of IscS (63). Third, the IscS dimer and the IscU dimer form a tight α₂β₂ complex, as shown using proteins from A. vinelandii and from E. coli, with a K_d of 2 μM (2, 141). Fourth, a covalent linkage, sensitive to dithiothreitol, is formed within this heterotetramer upon the addition of l-cysteine during cysteine desulfurase catalysis. This linkage is a disulfide bridge between Cys328 of IscS and Cys63 of IscU, suggesting that sulfur transfer is initiated by the attack of solvent-exposed Cys63 of IscU on the persulfide transiently produced by IscS (63).

Fig. 7Assembly of Fe-S clusters within IscU. The protein reacts with a cysteine desulfurase, an iron donor, and a reducing agent to assemble 2Fe-2S and 4Fe-4S clusters sequentially. e⁻, electron.

IscU from E. coli contains only three conserved cysteine residues, Cys37, Cys63, and Cys106, which are suggested to bind the clusters (40). Unfortunately, site-directed mutagenesis was not informative enough, and so far there is no three-dimensional structure of a cluster-containing form of IscU available for firmly supporting this view and for identifying a fourth ligand. During the incubation of holoforms of IscU from E. coli, A. vinelandii, and Thermotoga maritima with a variety of apoprotein targets (e.g., aconitase and ferredoxin), it was shown that 2Fe-2S and 4Fe-4S clusters can be efficiently and directly transferred (13, 84, 145). In general, the maturation of the target proteins through reaction with holo-IscU occurs at much higher rates than that through reaction with ferrous and sulfide ions. This observation provides strong support to the concept of IscU’s being a scaffold protein. A most interesting finding showing that aconitase, a 4Fe-4S enzyme, can be maturated only during reaction with the 4Fe-4S form of IscU and not with the 2Fe-2S one was recently reported (140). Whether this result reflects a univocal relationship between the type of cluster present on the scaffold and that eventually found in the target remains to be established.

IscA, an A-Type Fe-S Transporter

The effects of the inactivation of iscA are relatively minor compared to those observed with other isc knockouts. Nevertheless, an E. coli iscA mutant displays a reduced growth rate and reduced Fe-S enzyme activities (28, 139).

IscA is a homodimer able to bind both 2Fe-2S and 4Fe-4S clusters, probably by using the three conserved cysteine residues (Cys35, Cys99, and Cys101 for E. coli) during reaction with Fe²⁺, l-cysteine, and IscS (67, 102). Furthermore, it is able to deliver its clusters, intact and directly, to apoprotein targets in a reaction involving intermediate complexes (102, 103). This finding was interpreted to indicate that IscA has, like IscU, a scaffold protein function, even though the reason for the requirement of two different scaffold proteins encoded in the same operon is unclear. The fact that the clusters can be transferred from IscU to IscA whereas the reverse is not possible suggests a scenario in which the assembly of clusters is achieved exclusively within IscU and the maturation of a subgroup of Fe-S proteins does not result from a direct reaction with IscU but instead requires the involvement of IscA as an intermediate cluster carrier (103). The identity of these proteins and why they need a specific cluster donor different from IscU are unknown. There is no three-dimensional structure of a cluster-containing form of E. coli IscA. However, that of the 2Fe-2S-containing IscA from Thermosynechococcus elongatus (2.3-Å resolution) provides an interesting view of how this protein can bind a cluster (92). The cluster, partially solvent exposed, is coordinated asymmetrically by two conformationally distinct monomers, with the three conserved cysteine residues of one monomer and one cysteine of a second monomer providing the ligands. An alternative role for E. coli IscA was suggested by Ding and collaborators on the basis of the ability of IscA to bind ferric iron tightly (K_d, 3 × 10¹⁹ M⁻¹) in the absence of sulfur atoms (26, 27). It is proposed that IscA functions as an Fe storage protein from which, specifically, Fe atoms can be delivered to IscU. Recently, the same authors suggested a redundant activity for SufA to recruit iron and deliver it for the Fe-S cluster assembly process on IscU (83). However, the Fe-IscA/SufA species has been little characterized, and its physiological relevance remains to be established. In contrast, four examples of IscA/SufA proteins have been isolated with their cluster, either a 2Fe-2S or a 4Fe-4S cluster (91, 92, 150; our unpublished results).

HscA-HscB, a Chaperone-Cochaperone System

Adjacent to the iscRSUA genes are two genes, hscA and hscB, which encode members of the Hsp70 and Hsp40 chaperone families. The inactivation of hscA and hscB has effects on Fe-S enzyme activities of magnitudes similar to those of the effects of iscS mutation and is highly detrimental to the production of Fe-S proteins in the holoform (28, 139). As a matter of fact, the hscA gene had first been identified as being the site of a secondary mutation suppressing a subset of phenotypic defects due to mutation in the histone-like-protein-encoding hns gene (64). Consistently, HscA synthesis was subsequently shown to be induced by cold shock, like that of H-NS (74). A cold shock-responding promoter directing the expression of both hscB and hscA genes was identified, although no cis site for the cold shock CspA regulator could be found (74). Interestingly, this promoter site was mapped downstream of the iscA gene. This finding opens the possibility that under certain conditions, the synthesis of the chaperone-cochaperone pair may be uncoupled from that of the Isc proteins.

HscA is an ATPase, and HscB is its cochaperone (38, 143). IscU, in either the apo- or holoform, is a substrate of this chaperone system, as shown previously by the drastic stimulation of the ATPase activity of HscA upon incubation with IscU and HscB (50). HscA was shown to form a complex with IscU that is stabilized by both HscB and ADP and greatly destabilized by ATP (126). The C-terminal domain of HscA provides the substrate binding site, and the ATPase activities of variants lacking the C-terminal domain are unable to be stimulated by IscU. Remarkable biochemical studies by J. R. Cupp-Vickery, K. G. Hoff, and collaborators showed that the highly conserved LPPVK motif of IscU plays a major role in the IscU-HscA interaction and that the cochaperone HscB recognizes a different region of IscU (23, 49). A kinetic model of the HscA-ATPase cycle has been proposed (127). While mechanisms of IscU capture and release by the HscA-HscB chaperone-cochaperone system are well characterized, it is still unclear how this system participates in the Fe-S cluster assembly process. So far, the most consistent hypothesis is that HscA-HscB helps IscU to release its Fe-S cluster and to transfer it to apoproteins. By using proteins from A. vinelandii, it was shown previously that HscA-HscB indeed improves 2Fe-2S cluster transfer from IscU to apoferredoxin in an ATP-dependent reaction (18). Yet it was recently reported that IscU can transfer its 4Fe-4S cluster to apoaconitase in an HscA-HscB-independent mode (140). A possibility is that the HscA-HscB contribution is essential in vivo but dispensable under certain conditions in vitro.

Fdx, a Ferredoxin

In E. coli, fdx gene inactivation results in a reduced growth rate and reduced Fe-S enzyme activities (28, 139). The protein has been structurally characterized and shown to contain a stable 2Fe-2S cluster that is chelated by four cysteines and can exist under two redox states, the diamagnetic (2Fe-2S)²⁺ state and the paramagnetic reduced (2Fe-2S)¹⁺ one (62, 134). For that reason, this cluster is supposed to function as an electron transfer partner in the process of Fe-S cluster biogenesis, even though the physiological electron donor and acceptor partners are not known. The ferredoxin-NADPH oxidoreductase, the product of the fpr gene, is a potential donor, but in its absence, the activities of Fe-S enzymes such as dehydratases are not affected (28). There are several reactions during the process of cluster assembly in which electrons are required. First, whereas two sulfur atoms are introduced into the scaffold protein, IscU, for example, in the sulfane S⁰ state, they end up as two sulfide bridges once a (2Fe-2S)²⁺ cluster is assembled. This finding implies a two-electron reduction of each of the S⁰ atoms of the persulfides and, thus, a total amount of four electrons per assembled cluster. Only two electrons can be provided by the two Fe²⁺ atoms introduced into IscU which end up as Fe³⁺ ions, and a source of two additional electrons is required. So far, there is no evidence that Fdx participates in such a process. Second, electrons are required for the fusion of two (2Fe-2S)²⁺ clusters into a single (4Fe-4S)²⁺ cluster within the IscU protein (Fig. 7). Indeed, this reaction requires the injection of one electron into each 2Fe-2S cluster, and recent results may indicate that Fdx plays a role in this process (140).

THE Suf SYSTEM OF E.COLI AND SALMONELLA

Starting with a strain lacking the whole isc cluster (i.e., a ΔiscRSUA-hscBA-fdx-yfhJ::Km^r strain) which exhibits very strong and pleiotropic phenotypes such as auxotrophy towards nicotinic acid, thiamine, methionine, isoleucine, and valine, Takahashi and Tokumoto searched for extragenic suppressor mutations (136). Likewise, spontaneous nicotinate-independent pseudorevertants were isolated on a medium including all other requirements. Five such suppressor strains were isolated, all of them having mutation in the promoter region of the suf operon (136). The suf operon is made of six genes in the sufABCDSE order. The lack of a functional suf operon is neutral for E. coli under normal growth conditions (96, 108, 136). In contrast, under oxidative stress, the deletion of the suf genes makes E. coli unable to produce functional forms of enzymes containing oxygen-labile Fe-S clusters (96). The same observation was obtained when cells were exposed to 2,2'-dipyridyl, an iron chelator (96, 108). An E. coli strain lacking both functional isc and suf operons is not viable, implying that at least one essential Fe-S protein needs either one of the two systems for its maturation, regardless of the environmental conditions (108, 136).

SufSE, the Cysteine Desulfurase

SufS (previously CsdB) is a PLP-dependent cysteine desulfurase, but the level of its activity in vitro has been shown to be much lower than that of IscS (89, 90). However, SufS binds tightly to SufE, and the resulting 1:1 complex, observed both in vivo and in vitro, displays much greater activity than SufS alone, making this new type of heterodimeric cysteine desulfurase as efficient as IscS (81). Site-directed mutagenesis showed that the conserved cysteine (Cys51) of SufE is critical for this stimulatory effect (81). Thus, the homodimeric SufS and SufE proteins work together to provide sulfur atoms to the Suf biosynthetic machinery. Sulfur enters at SufS, in the form of l-cysteine, and comes out at SufE, in the form of a persulfide. This process is possible due to the generation of a persulfide at Cys364 of SufS through cysteine desulfuration and to the conversion of Cys51 of SufE into a persulfide, which serves as a source of sulfane S⁰ atoms, through a transpersulfuration reaction. Sulfur atom transfer from SufS to SufE was unambiguously demonstrated using ³⁵S labeling experiments and mass spectrometry (101, 109).

SufS, under different forms, and SufE have been structurally characterized, providing strong molecular support to the proposed cysteine desulfurase mechanism. In particular, a three-dimensional structure of SufS with its persulfide at Cys364 is available (78). Furthermore, these structural studies show that the catalytically essential Cys364 is located in a deep cleft, 5 Å away from PLP, in a region of the polypeptide chain with limited flexibility (78). This finding may explain why the activity is so weak and why the limiting step of the reaction is the formation of the persulfide at Cys364 (137). The structure of SufE shows that the sulfur-accepting residue Cys51 is located at the top of a loop that is buried in a hydrophobic cavity, not easily accessible (40). Interestingly, the SufE structure resembles that of IscU, strengthening the notion that the two proteins share the property of acting as acceptors of sulfur atoms provided by cysteine desulfurases (118). Thus, in the absence of a SufS-SufE three-dimensional structure, it is difficult to understand how the SufS-to-SufE sulfur transfer reaction at the molecular level occurs and the origin of the stimulating effects of SufE on the SufS cysteine desulfurase activity.

SufA, an A-Type Fe-S Transporter

E. coli SufA has 44% homology to E. coli IscA and shares with IscA the ability to bind 2Fe-2S and 4Fe-4S clusters (105). SufA is a true Fe-S protein, as shown by the observation that it can be isolated from E. coli cells in a form containing a 2Fe-2S cluster, characterized by Mössbauer spectroscopy and Raman resonance spectroscopy (our unpublished data). The clusters can be transferred in vitro directly and efficiently from SufA to apoprotein targets, such as ferredoxin and biotin synthase (103, 105). The maturation of these proteins occurs at much higher rates when iron and sulfur are provided by holo-SufA than when they are provided by ferrous sulfate and sodium sulfide salts. Furthermore, these reactions are not inhibited by iron chelators and proceed via detectable complexes between SufA and the cluster acceptor proteins (103). This finding is consistent with a direct and concerted interprotein cluster transfer, with no liberation of iron and sulfur in solution. The assembly of Fe-S clusters within SufA has been extensively studied. The transfer of sulfur atoms from SufE to the conserved cysteines of SufA (Cys50, Cys114, and Cys116 for E. coli), supposed to act as cluster ligands, through a transpersulfuration reaction has been observed previously by mass spectrometry and peptide-mapping experiments (123). Furthermore, the addition of ferrous iron in the presence of a reducing agent to the sulfurated form of SufA results in the formation of protein-bound Fe-S clusters (123). These results are consistent with the view that SufA acts as a scaffold protein in the Suf machinery. However, another possibility is that SufA, as suggested for IscA, is an intermediate cluster carrier which gets its clusters from IscU, for example, or from other proteins of the Suf system, such as SufB (see below), and delivers them to a specific target(s).

SufBCD, an ATP-Hydrolyzing Component

Strong phenotypes of sufB, sufC, and sufD mutants of several living organisms, including plants, have been observed (48, 51, 146). Detailed studies with the plant pathogen enterobacterium Erwinia chrysanthemi have been carried out and have shown specific roles of the three proteins in siderophore utilization and, by extension, in virulence (96).

ABC complexes form a very well known class of ubiquitous cellular machineries catalyzing the import and export of various solutes across membranes. SufC contains all motifs that are characteristic of the ABC ATPases, like the Walker site A and B motifs, as well as motif C (79), and is endowed with an ATPase activity (95, 96). Two-hybrid, copurification, and tandem affinity purification tag methods have demonstrated that SufB, SufC, and SufD form a tight complex (96, 109). In addition, exclusion chromatography and mass spectrometry have allowed the definition of a SufBC₂D stoichiometry (G. Layer et al., unpublished results). The complex is located in the cytosol of E. coli, somehow in agreement with the absence of predicted transmembrane segments in any one of the three partner sequences. So far, there is no clear interpretation for the presence of such a cytosolic pseudo-ABC transporter in the Suf system. Fe-S cluster assembly may require energy, and ATP hydrolysis by the SufBC₂D complex is likely to provide it. Alternatively, as suggested by Outten et al., SufBC₂D may provide a transport and channelling function for sulfide, thus preventing sulfide leakage (109). This hypothesis is supported by the recent observations that (i) SufBC₂D makes a complex with SufSE through an interaction between SufE and SufB (K_d, 2.7 μM) (70); (ii) SufB can trap the sulfur atoms of SufE, resulting in the formation of persulfides and polysulfides on SufB (70); and (iii) SufBC₂D stimulates the cysteine desulfurase activity of SufSE, although in a surprisingly ATP-independent manner (70, 109).

The roles of SufB and SufD in Fe-S cluster assembly are unknown so far. SufD from E. coli, whose sequence shows no known predicted motifs, has been structurally characterized and was shown to be free of any cofactor or prosthetic group (5). In contrast, SufB can assemble a 4Fe-4S cluster which can interconvert between +2 and +1 redox states, as shown by Mössbauer and electronic paramagnetic resonance spectroscopy (70). The function of this cluster is unknown. It may participate in a redox process, perhaps like Fdx in the Isc system. Alternatively, SufB may be the scaffold protein of the Suf system, receiving the sulfur atoms from SufE and then assembling a 4Fe-4S cluster which, subsequently, may be transferred to apoprotein targets.

REGULATORY ASPECTS

Regulation of the isc Operon

The Isc operon includes its own regulator, IscR, which is encoded by the promoter-proximal gene (121). IscR exhibits aminoacid sequence similarity to the Mar/Rob/Sox family of transcriptional regulators. Electron paramagnetic resonance studies have shown that, as isolated, IscR protein contains a 2Fe-2S cluster (121). Deleting iscR yields increased expression of the downstream iscSUA-hscBA-fdx genes (39, 121, 149). Mutations in genes encoding the Isc system lead to the increased expression of the isc operon, indicating that it is the 2Fe-2S holoform of IscR that acts as a repressor and that the IscR protein is maturated by the Isc system (121). Also, mutations of the cysteine residues thought to act as ligands for the Fe-S cluster result in the increased expression of the isc operon (149). This finding led to an attractive model based on a feedback mechanism (Fig. 8). When cells are able to make Fe-S clusters, the IscR protein is in its holoform and shuts off the Isc machinery. In contrast, unfavorable conditions, e.g., a shortage of elementary components or oxidative stress, promote the occurrence of IscR in its apoform, which leads to the derepression of the isc genes and, hence, to the reinitiation of the assembly machinery. The cycle is closed when the apo-IscR is converted back into its repressing holoform. That exposure to oxidative agents, iron chelators, or conditions of sulfur shortage leads to the iscR-dependent induction of the isc operon has been experimentally confirmed (43, 108, 149).

Fig. 8Schematic representation of the regulation of the isc and suf operons. Transcriptional regulators that control the expression of the isc and suf operons are indicated. The regions protected by IscR (hatched rectangles), Fur (black rectangle), IHF (white rectangle), and OxyR (grey rectangle) in DNase I footprinting assays are indicated (39, 108, 149). The boundaries of the binding regions are indicated by the positions of the nucleotides relative to the transcription start site. The transcription start site (+1) determined by primer extension is indicated by an arrow.

Regulation of the suf Operon

The suf operon is the target of multiple trans-acting factors (Fig. 8). First, the Fur (ferric uptake repressor) protein represses the suf expression under conditions of iron repletion (95, 108, 113). It is interesting that two of the mutations that suppress the isc defect alter the binding site for the Fur repressor (136). Second, the OxyR activator allows hydrogen peroxide-mediated induction (73, 154, 155). Third, the integration host factor (IHF) facilitates contact between RNA polymerase and OxyR (73). Interestingly, the transposition of an IS1 element at position −149 (relative to the transcription start site) in the IHF binding site allows the suppression of the isc mutation (136). Fourth, IscR is likely to mediate the induction of suf under Fe-S cluster-damaging conditions, such as a shortage of iron and oxidative stress (39, 73, 149). It is important to underscore that it is the apoform of IscR that acts as a transcriptional activator (Fig. 8). This idea is supported by the fact that an IscR mutant lacking all three Cys residues is still able to activate suf expression (149). All of these four regulators make direct contact with the promoter region of the suf operon, and well-defined cis-acting sequences have been described previously (Fig. 8). Thus, the oxidative stress induction of the suf operon is mediated via two independent systems, OxyR and IscR, which appear to function in an additive manner. Indeed, significant induction in an oxyR or an iscR mutant was observed, while no induction was apparent when both genes were deleted (149). The same applies for iron starvation that is mediated both by apo-IscR activation and Fur derepression. Altogether, these findings led to the view that the Suf machinery is the stress-responsive system (see below for a discussion).

OTHER PROTEINS INVOLVED IN IRON-SULFUR METABOLISM

Several other factors have been suggested to participate in cluster assembly and repair in E. coli and Salmonella. Most of them were identified by their abilities to act as extragenic and/or multicopy suppressors of mutations in Fe-S cluster metabolism, while others possess biochemical properties that are consistent with a role in Fe-S cluster biogenesis. The question of their relationship with the Isc and Suf machineries remains open.

CsdA-CsdE, Another Cysteine Desulfurase

Besides IscS and SufS, CsdA is the third PLP-dependent cysteine desulfurase enzyme in E. coli. CsdA shows 45% identity to SufS, and its activity depends on a catalytic cysteine residue (Cys358) which is the site of persulfide formation (89). The following observations support the notion that CsdA has a specific sulfur acceptor partner, named CsdE, generating a two-component cysteine desulfurase system comparable to SufS-SufE. First, the corresponding gene csdA lies immediately upstream of csdE (previously known as ygdK), which was shown to have 35% identity to SufE, and three-dimensional structures of CsdE and SufE proteins reveal striking similarity (40, 118). Second, CsdA forms a tight complex with CsdE, which can be observed both in vivo and in vitro (82). Third, CsdE stimulates CsdA cysteine desulfurase activity, and specific sulfur transfer from Cys358 of CsdA to Cys61 of CsdE occurs through a transpersulfuration reaction, as shown by mass spectrometry and site-directed mutagenesis (82).

The participation of the CsdA-CsdE system in Fe-S cluster assembly is also suggested by the following in vivo and in vitro results. First, the csdA and csdE genes can act as multicopy suppressors of deficiencies in NAD, thiamine, and Fe-S enzyme activities in iscS mutant strains (82). Second, the l-cysteine-CsdA-CsdE system can be used as a source of sulfur atoms for the assembly of clusters in vitro (82). In fact, it has been proposed that CsdA-CsdE may have a specific function in the maturation of NadA, the quinolinate synthase, an Fe-S enzyme catalyzing a critical step in NAD synthesis (20, 82, 104). Indeed, the csdA-csdE pair behaves as a multicopy suppressor of NAD auxotrophy associated with an iscS mutation, whereas csdA alone does not (82). Furthermore, CsdA-CsdE is more efficient than IscS and SufS-SufE in promoting Fe-S cluster assembly within NadA in vitro. Finally, the observation that the quinolinate synthase of Arabidopsis thaliana is made from a fusion of a CsdE-like domain with a NadA domain nicely illustrates the privileged relationship between CsdE and NadA (40, 82, 94).

ErpA, an A-Type Fe-S Transporter Essential for Respiration

In E. coli, a new A-type Fe-S protein, which we named ErpA (previously known as YadR), was identified previously. The erpA gene was found to be repressed by IscR under anaerobiosis (39). The ErpA primary sequence exhibits 40% identity to the sequence of IscA and 34% identity to that of SufA from E. coli. This finding suggested that ErpA may fulfil a role similar to that of IscA or SufA, as a scaffold protein or Fe-S transporter, in Fe-S cluster biogenesis. The results of in vitro investigations were in line with this expectation, since ErpA was indeed found to allow the maturation of several Fe-S enzymes. The unexpected arose with attempts to inactivate the chromosomal erpA copy. Indeed, the invalidation of the chromosomal wild-type allele was invariably accompanied by the duplication of a large region, yielding a partially diploid strain containing both the wild-type and the mutated alleles, suggesting that erpA is an essential gene (80). The use of a conditionally expressed allele allowed us to show that erpA is indeed essential under respiratory conditions, i.e., in the presence of oxygen or alternative electron acceptors such as nitrate or fumarate (80). Accordingly, a null allele can be obtained by maintaining the cells under fermentative growth conditions. IspG and IspH proteins were considered to be good candidates among essential Fe-S proteins that would be required in any respiratory metabolism. IspG and IspH are two essential 4Fe-4S enzymes responsible for the synthesis of isopentenyl diphosphate, a precursor of quinones, the electron carriers present in all types of respiratory chains (122). The ectopic expression of a eukaryotic isopentenyl phosphate biosynthesis pathway allows the elimination of the respiratory defects of an erpA mutant (80). Also, the amounts of ubi- and menaquinone in the erpA mutant are greatly reduced compared to those in the wild type. It was thus concluded that isoprenoid biosynthesis and, more specifically, the activities of IspG and IspH are specifically affected by the erpA mutation. However, it is fair to say that the bypassing of the IspG-IspH-catalyzed steps by the eukaryotic pathway does not fully restore the wild-type phenotype, and it is very likely that ErpA controls the maturation of an as-yet-unknown additional set of Fe-S proteins.

NfuA, a Fusion of Two Scaffolds?

The product of the nfuA gene (formerly known as yhgI) of E. coli has been recently identified as a novel Fe-S cluster binding protein (3). It has been named NfuA because its C-terminal domain exhibits homology to the C-terminal cluster binding Nfu domain of the A. vinelandii scaffold protein NifU (30) and because the N-terminal domain is homologous to IscA/SufA, however, without the cysteines thought to act as cluster ligands in the latter. Site-directed mutagenesis showed previously that both cysteines of the NfuA conserved CXXC motif of the C-terminal domain are functionally important in vivo (3). The purified protein binds a 4Fe-4S cluster at the interface of two monomers, likely coordinated by the CXXC motif of each monomer, which can be efficiently transferred to apoaconitase (3). Similar results were obtained with the NfuA protein from A. vinelandii (6).

An E. coli strain lacking a functional nfuA gene exhibits increased sensitivity to oxidative agents, cobalt exposure, and iron starvation (3). Moreover, additive phenotypic defects are observed when nfuA and isc or suf mutations are combined. The nfuA gene was found to be repressed by IscR, mainly under anaerobiosis (39). These observations support the hypothesis that NfuA acts as a scaffold protein during cluster assembly or as an [Fe-S] cluster-trafficking protein or participates in [Fe-S] protein folding and repair under stress conditions. In support of the latter possibility is the fact that nfuA was repeatedly found in several transcriptomic analyses investigating the response of E. coli to protein misfolding-related stress, including the heat shock response (76, 99, 124). Hence, the hypothesis that NfuA may act both as a maturase and as a chaperone constitutes an exciting idea to pursue.

Potential Sources of Iron

Iron salts are most often used as a source of Fe atoms for Fe-S cluster assembly in vitro using purified proteins. However, it is generally assumed that such high concentrations of free iron are not available within cells because of the toxicity associated with the Fe-dependent production of oxygen radicals (the Fenton reaction). Thus, it is believed that cells contain specialized proteins for the binding and trafficking of iron and thus limit free and toxic forms of iron. This issue has been the focus of a great number of studies and interpretations. However, so far there is no clear idea about the number of these Fe-buffering systems and their specificities. For example, it is still unknown whether there is a specific Fe donor for Fe-S cluster assembly.

CyaY, the Bacterial Frataxin.

Frataxin was first identified as the missing protein in patients with Friedreich’s ataxia, a progressive cardio- and neurodegenerative disease resulting from abnormal Fe homeostasis and oxidative damage (4, 17, 45). The disruption of the frataxin-encoding gene, YFH1, in yeast results in the accumulation of iron in mitochondria and deficiency in Fe-S enzyme activities (31, 36, 85, 93).

E. coli and Salmonella contain a frataxin homolog, namely, the product of the cyaY gene. The E. coli gene was shown to complement the loss of Fe-S enzyme activities and respiratory function and to eliminate susceptibility to oxidative stress in a yeast strain lacking a functional YFH1 (10). On the other hand, the deletion of cyaY alone in E. coli and Salmonella does not cause any obvious alteration of the Fe balance and confers no phenotype (77, 144). This finding probably reflects the functional redundancy of CyaY and other bacterial proteins. For example, a double knockout of cyaY and yggX (see below) in Salmonella results in defects in Fe-S enzyme activities and increased sensitivity to paraquat (144). The purified CyaY protein from E. coli binds iron tightly (K_d > 10¹⁷ M⁻¹), mainly in the ferric form, and can deliver it to the scaffold protein IscU to generate a 2Fe-2S cluster in the presence of l-cysteine and IscS (71). Interestingly, CyaY was shown to form a transient complex with IscS, which interacts with IscU (71). Together, these observations establish an obvious link between CyaY and Fe-S cluster biosynthesis.

YggX, an Fe-Chelating Antioxidant Protein?

The yggX gene encodes a small and abundant 11-kDa protein and is found in eubacterial genomes but not in archaeal and eukaryotic ones. It was first identified in Salmonella as an extragenic suppressor of thiamine auxotrophy exhibited by a glutathione-deficient gshA mutant (41). Surprisingly, the suppressing effect was due to a promoter mutation that increased the level of yggX expression. Subsequently, it was shown that yggX overexpression can also suppress phenotypes associated with isc mutations, such as the requirement for thiamine and nicotinic acid and decreased Fe-S enzyme activities (130, 131). Studies of E. coli showed that deleting yggX had no effect on aconitase activity under normal growth conditions. In contrast, under oxidative stress induced by paraquat, the yggX mutant exhibited a much larger drop in aconitase activity than the wild type (117). Moreover, in the presence of paraquat, the overexpression of YggX yielded a higher-than-normal level of aconitase activity. These data demonstrated that YggX plays an antioxidant role in the cells and can afford protection against oxidative damage to Fe-S proteins.

There are several circumstantial observations that tie together yggX, oxidative stress, Fenton chemistry, and mutagenesis control. In particular, yggX belongs to a four-gene complex operon, wherein it lies in the second position downstream from the mutY gene (41). MutY is a glycosylase that removes A from an A-oxoG mismatch, hence participating in the repair of oxidatively damaged DNA. Moreover, it contains an Fe-S cluster, the role of which may be redox sensing of base lesions (12, 14, 148). It was therefore tempting to picture YggX as being dedicated to the maturation of the MutY protein under oxidative stress. The overexpression of YggX was indeed found to decrease the spontaneous mutation frequency of cells by reducing the damage to DNA by hydroxyl radicals (42). Also, in vitro, purified YggX was found to protect DNA from Fenton chemistry. However, no activating or protecting effect of YggX protein on MutY activity could be found (42). Curiously also, whereas the yggX gene belongs to the superoxide-sensing SoxR-SoxS regulon, the mutY gene does not, because the regulated yggX promoter is located internally in the operon, downstream of the mutY coding sequence (117).

The current model is that YggX is a component of the Fe-trafficking machinery, which limits the oxidation of Fe-S clusters or facilitates their repair through the sequestration of cellular Fe(II), making it available for carefully controlled Fe-dependent cellular processes while removing it from involvement in Fenton chemistry (42). However, based on findings from biochemical studies of E. coli and Salmonella, there is still no clear evidence that YggX binds Fe specifically, and either specific conditions or missing partners have been put forward to account for the lack of or weak iron binding observed in vitro (22, 107).

YtfE, a Nonheme Di-Iron Protein.

An E. coli strain lacking an active ytfE gene displays increased sensitivity to oxidative stress, nitric oxide, and iron starvation (59, 60, 61). The expression of the ytfE gene is under the control of the Fnr (fumarate-nitrate reductase regulator), NsrR (nitrite-responsive repressor), and Fur transcriptional regulators (32, 59). More interestingly, in the ytfE mutant, a number of Fe-S enzymes such as nitrate reductase, glutamate synthase, and aconitase have decreased activity levels (59, 60). The inactivation of these enzymes by hydrogen peroxide or nitric oxide occurs at much higher rates in the mutant than in the E. coli wild-type strain (59, 60). It seems that YtfE plays an important role specifically in the repair of Fe-S clusters damaged by oxidative and nitrosative stress conditions. This possibility is supported by the observations that the inactivation of ytfE abolishes the recovery of Fe-S enzyme activities once the stress is scavenged and that the addition of the YtfE protein to extracts of cells exposed to hydrogen peroxide results in the full recovery of fumarase A and aconitase B activities (60). Since it is a nonheme di-iron protein, an interesting hypothesis is that YtfE is involved in Fe mobilization during cluster repair. If proven, that would be the first example of a nonheme binuclear iron protein’s being involved in Fe-S metabolism.

Ferritin B.

Salmonella contains four ferritins, FtnA, Bfr, Dps, and FtnB (142). A function for FtnB as an iron donor for Fe-S cluster repair was suggested previously, and noticeably, an ftnB mutant is affected in virulence (142). The E. coli genome also contains an ftnB gene, but so far its implication in Fe-S metabolism has not been investigated.

Role in Fe-S Biogenesis for the yfhJ-pepB-sseB Genes?

Three genes, yfhJ, pepB, and sseB, are located immediately downstream from the iscRSUA-hscBA-fdx genes. Since the distance between fdx and yfhJ is 11 nucleotides, it is likely that all these genes belong to the same transcriptional unit. Moreover, they were found previously to be part of the IscR regulon (39). Therefore, the question arises as to whether these three genes also participate in Fe-S cluster biogenesis. Available information on each suggests that this possibility may be so. (i) YfhJ, also referred to as IscX, is a small acidic protein that has been structurally characterized both in solution (by nuclear magnetic resonance spectroscopy) and in crystal form (by X-ray crystallography) (112, 125). IscX displays some affinity for IscS, and an IscX-IscS complex has been observed (112, 138). Furthermore, IscX seems to be able to bind Fe atoms weakly at a surface involved in IscS binding and containing glutamate and aspartate residues (112). (ii) PepB is an aminopeptidase that has been shown by a global proteomic approach to interact with HscA (16). (iii) SseB overexpression confers the elusive serine sensitivity phenotype, which may be related to the metabolism of l-cysteine, the source of sulfur for Fe-S cluster assembly (44).

ApbC, Another ATPase

The apbC gene was found previously to participate in thiamine biosynthesis in Salmonella, possibly as a consequence of the ApbC protein’s playing a role in the maturation of ThiH (114), an enzyme of the radical-SAM family with an oxygen-sensitive Fe-S cluster, catalyzing thiazole synthesis. The notion that this protein contributes to the assembly, protection, and/or repair of Fe-S clusters in general is supported by the following observations (115, 129): (i) the phenotypes with regard to thiamine synthesis caused by mutations in apbC are similar to those found for mutants defective in isc; (ii) aconitase and succinate dehydrogenase have decreased specific activities in the apbC mutant compared to those in the wild type, although to a lesser extent than those in the isc mutant; and (iii) isc and apbC mutations exert cumulative effects with respect to these Fe-S enzyme activities. It should be noted that these phenotypes were observed in a yggX mutant background (129). Homologs of apbC have been found in E. coli (mrp) and in yeast (NBP35 and CFD1) (47, 119, 147). In yeast, they are proposed to participate in Fe-S cluster assembly in the cytosol. ApbC has been purified to homogeneity and shown to display an ATPase activity in vitro, in line with the presence of a functionally important Walker P-loop motif indicative of a nucleotide binding protein (129). As discussed above, the Isc and Suf systems also contain proteins with ATPase activities, such as HscA and SufC, respectively, and in general, metal cofactor assembly machineries include proteins with nucleotidase activities (21, 46, 72, 87). Whether ApbC functions to facilitate ATP-dependent Fe insertion into Fe-S clusters during the assembly of clusters or cluster transfer remains to be studied. The same authors who identified ApbC identified another potential protein involved in Fe-S metabolism, namely, the ApbE protein (8, 129). The apbE mutant displays a phenotype similar to that of the apbC mutant. The ApbE protein was shown previously to be a periplasmic lipoprotein whose periplasmic location but not membrane association is required for efficient thiamine biosynthesis (8, 9). The ApbE sequence displays no known predicted motif, and the function of the protein is not known yet.

CONCLUSIONS

In the last decade, identifying and unravelling the functioning of cellular pathways assisting Fe-S cluster biogenesis have become key issues in several areas of research in biology, including microbial pathogenicity, molecular medicine, structural genomics, plant development, and aging. In this context, E. coli has been highly cooperative as a model system for investigating processes that in most instances are well conserved throughout all living organisms. A remarkable synergy between genetics and biophysics, along with inorganic chemistry and molecular biology, has made possible the identification of several factors necessary for Fe-S cluster biogenesis. Although this field is quite recent (it began after the seminal paper of Jacobson and colleagues on nitrogenase in 1989) (53), in retrospect, it is compelling to see how classic genetics has been instrumental and efficient in tackling the Fe-S cluster biogenesis issue. Reading the related literature is like flicking through a textbook of classic genetics: hunting for extragenic and multicopy suppressors to identify new components; combining mutations, including lethal synthetic ones, to investigate additive effects; and isolating conditional mutations—all of these approaches have been used successfully and have been highly rewarding. One of the reasons for the Fe-S cluster biogenesis area to have been a playground for the geneticist likely relates to the fact that redundancy is at the heart of this field. Indeed, owing to the pivotal roles of the Fe-S proteins in cellular processes, cells have taken the wise, though noneconomical, path of developing multiple systems for ensuring the same functions, namely, Fe-S cluster biogenesis and insertion into apoproteins. The resulting current picture, at least for E. coli and Salmonella, is that there are two hard-core systems, the Isc and Suf systems, whose actions are supplemented or paralleled by a series of ancillary factors (Fig. 2).

A main issue regarding the Isc and Suf systems concerns their specificities. First, one may ask whether they maturate the same set of apoproteins. The fact that an iscS sufS double mutant is nonviable indicates that the Isc and Suf systems share at least one essential Fe-S apotarget. Yet suf mutants are hypersensitive to Fe starvation, whereas isc mutants are not, and such hypersensitivity is not suppressed by the overproduction of the isc genes. This finding points to the occurrence of specific links between the Suf system and iron metabolism, thereby implying a certain level of specificity as far as the Suf and Isc systems are concerned. Second, one may ask whether these systems operate under the same conditions. The answer seems to be no, but the issue is complex and deserves careful examination. Let us consider, for instance, the situation when cells are under aerobic conditions that are well known to be suboptimal for Fe-S cluster biogenesis: the IscR protein will continuously oscillate from its holoform, sensitive to oxygen, to its apoform, which allows its conversion back to the holoform via the feedback control that the apoform exerts on Isc synthesis (Fig. 8). In this situation, it is very unlikely that the Suf system would be fully activated since it would also require OxyR-mediated activation and Fur derepression. Hence, one can assume that under aerobiosis, Fe-S cluster biogenesis is carried out essentially by the Isc system. Let us now consider conditions that are actually detrimental for Fe-S biogenesis, such as oxidative stress and/or a shortage of iron or sulfur. IscR will occur mainly in its apoform, and the Isc system will be synthesized, while concomitantly, apo-IscR- and OxyR-mediated activation and Fur derepression will allow the full synthesis of the Suf system. The biogenesis of Fe-S proteins in this case relies on both Isc and Suf systems. Although the same drug (paraquat, or 2,2'-dipyridyl) induces isc and suf gene expression, it is probably erroneous to assume that the Isc and Suf systems operate under the same conditions. Reciprocally, the view that the Isc and Suf systems are used under favorable and unfavorable conditions, respectively, is probably too simplistic. Rather, E. coli and Salmonella may satisfy a continuous Fe-S cluster biogenesis demand by coupling the synthesis of both systems to evolving cellular or environmental conditions. Moreover, since the systems are connected via the IscR regulator, the Suf system can be activated when the Isc system is overwhelmed by the demand. If the model described above holds, it is predicted that the Suf system operates under extremely unfavorable conditions. Identifying the intrinsic molecular features of Suf proteins that make them more efficient than Isc proteins in getting iron and/or sulfur under limiting conditions or in functioning in the presence of increased concentrations of free radicals will constitute a very important question to tackle to understand each machinery’s contribution to Fe-S biogenesis.

Another issue regarding the Isc and Suf systems concerns the specific functions of each Suf and Isc protein within each system. Noticeably, the two systems share only two structurally and biochemically related compounds: the cysteine desulfurases IscS and SufS, which provide sulfur, and IscA/SufA, for which the best prediction of function is as Fe-S cluster carriers. The role of the SufBCD pseudo-ABC transporter remains uncertain. It has been proposed to function as an intracytosolic transporter acting as a channel to control reactive species such as sulfide and ferric ion. Its partnership with SufS indicates that this hypothesis remains valid. However, the role of the ATPase is enigmatic. Also, the presence of an Fe-S cluster within SufB suggests that it may act as the scaffold of the Suf system. Such a hypothesis is strengthened by the fact that no homologs of IscU, the well-characterized and indisputable scaffold of the Isc system, are present in the Suf system. It is striking that were this scaffold hypothesis for SufB be confirmed, one would face a fascinating case of convergent evolution between the two machineries. Indeed, both systems would have evolved a subcomplex made of a cysteine desulfurase (IscS or SufSE) interacting with a scaffold (IscU or SufB) and an ATP-hydrolyzing component (HscBA or SufC). The coupling of such a complex with an A-type carrier may set the minimal requirements for Fe-S cluster biogenesis.

Then there are the non-Suf and non-Isc ancillary factors. The definition of the biochemical functions of these ancillary proteins, as well as the analysis of the phenotypic consequences of their loss, is an immediate step that many groups are currently taking. Another important challenge will be to know whether these components transiently associate with the Isc and/or Suf machineries to define a series of functional pathways. Alternatively, they may constitute bona fide new Fe-S cluster biogenesis pathways.

A crucial question that emerged along with the growing list of Fe-S cluster biogenesis factors relates to the notion of the repair of damaged Fe-S clusters. Indeed, Suf, NfuA, YggX, YtfE, ErpA, and ApbC all have been proposed to fulfil a repair function. As a matter of fact, the question of whether repair occurs has received little experimental attention (28). So the question arises of whether damaged Fe-S clusters are actually repaired and, if so, which components are doing the job. Are the Suf and Isc systems instrumental in the repair? Conversely, if the damaged Fe-S proteins are simply degraded and replaced de novo, how is the proteolytic cascade regulated? Which proteases are used? How do they sense their damaged substrates? Studying the fate of the damaged substrates is likely to provide us with a new, exciting facet of Fe-S protein dynamics.

Over the last decade, an unexpectedly large number of factors and potential pathways to assist Fe-S cluster biogenesis have been described. Reciprocally, the next question to address concerns the size and diversity of the set of Fe-S proteins waiting to be matured. How many Fe-S proteins are out there? A survey of highly reliable cases led to the number 88, but by using different criteria, a reasonable bet is to predict approximately 200 Fe-S proteins in E. coli (33). Fe-S proteins can harbor different types of clusters. Is there a specific maturation pathway for each of them? Some clusters will be buried inside the target structure, others exposed. Are there different ways, relying upon different components, for inserting buried versus exposed clusters? A related issue is whether Fe-S clusters are inserted in a cotranslational versus a posttranslation mode.

As mentioned in the introduction, their versatile redox chemical properties have largely contributed to the selection of Fe-S clusters by living organisms. Another reason for their successful use that has been repeatedly put forward is that iron and sulfur were in large abundance in the first billions of years of life. This suggestion has led to the view that Fe-S clusters can be spontaneously assembled by exploiting available resources. As recently argued by Imlay, the oxygenation of the earth gradually made the use of Fe-S proteins more and more difficult owing to iron limitation and the destabilization of their clusters in an oxidizing environment (52). Accordingly, cells have evolved sophisticated molecular devices to form, insert, and possibly protect Fe-S proteins. We have described how E. coli (and Salmonella) has built two hard-core systems, besides which numerous additional elements come in, to carry on Fe-S cluster biogenesis. We can already guess from their genomic contents that other bacterial species have evolved different strategies. For instance, Mycobacterium species have retained only the Suf system, whereas Acinetobacter uses the Isc system only and Bacillus subtilis harbors mixed Isc-Suf machinery. Present research in our laboratories aims at analyzing the biodiversity of these cluster biogenesis systems, as well as their history, particularly in relation to the context of the anaerobic-aerobic life switch.