Fe-S Proteins that Regulate Gene Expression

Abstract

Iron-sulfur (Fe-S) cluster containing proteins that regulate gene expression are present in most organisms. The innate chemistry of their Fe-S cofactors makes these regulatory proteins ideal for sensing environmental signals, such as gases (e.g. O₂ and NO), levels of Fe and Fe-S clusters, reactive oxygen species, and redox cycling compounds, to subsequently mediate an adaptive response. Here we review the recent findings that have provided invaluable insight into the mechanism and function of these highly significant Fe-S regulatory proteins.

Introduction

Given their vital function in controlling transcriptional and post-transcriptional processes, regulatory proteins play a large role in the central dogma of biology. In the case of bacteria, many regulatory proteins coordinate gene expression in response to specific environmental cues, which are often sensed through their co-factors. Due to the versatile chemical reactivity of iron-sulfur (Fe-S) clusters, several regulatory proteins have evolved to exploit this property and employ Fe-S clusters as co-factors, enabling prokaryotes, in addition to eukaryotes, to rapidly sense and adapt to their surroundings. Here, we aim to give a general overview of known or putative Fe-S regulatory proteins that modulate gene expression, emphasizing the recent accomplishments in this area of study.

Widespread in nature, Fe-S proteins participate not only in gene regulation, but have prominent functions in several diverse biological processes, including respiration, photosynthesis, nitrogen fixation, RNA modification, and DNA replication and repair. The activities of many enzymes rely on their Fe-S clusters, as these prosthetic groups can play structural, catalytic or electron transfer roles [1, 2]. These same intrinsic properties make Fe-S clusters ideal as cofactors for proteins that control gene expression [3–6]. Typically, Fe-S regulatory proteins contain [4Fe-4S] or [2Fe-2S] clusters in the oxidized (2+) or reduced (1+) redox state. Because of their capacity to delocalize electrons over both Fe and S ions, Fe-S clusters are susceptible to oxidation from molecules like redox-cycling drugs, oxygen (O₂), and reactive O₂ species (ROS), such as superoxide (O₂⁻) and hydrogen peroxide (H₂O₂), potentially leading to cluster conversion or complete cluster loss [7–9]. Fe-S clusters are also vulnerable to damage by nitric oxide (NO), which forms various Fe-nitrosyl species [10]. These signal-induced alterations of Fe-S clusters may propagate a conformational change within the regulatory protein such that its activity is altered. In addition, some regulatory proteins use the presence or absence of their own Fe-S cluster as a read-out in assimilating fluctuations in cellular levels of Fe and Fe-S clusters. Thus, Fe-S clusters enable regulatory proteins to sense one or more of these signals, thereby modifying protein function, and subsequently, the expression of target genes.

Fe-S clusters also exhibit variable sensitivities to signaling molecules, depending on the protein environment surrounding the cluster (for examples, see references [11–15]). Although we do not yet understand fully the basis behind this observation, some rules are emerging. For example, Fe-S clusters may be sheltered within the protein structure or are positioned such that they are readily solvent-exposed. The latter is predicted to be the case for most Fe-S regulatory proteins to efficiently carry out their sensing function; however, this has only been shown definitely for SoxR and IRP1 (see text below and Fig. 1A) [16, 17]. Additionally, while the iron atoms of [4Fe-4S] and [2Fe-2S] clusters are typically bound by four cysteine residues, other residues (e.g., histidine, aspartate), and glutathione may also serve as cluster ligands [18]. Redox potentials of Fe-S clusters can also vary, gearing Fe-S regulatory proteins toward sensing specific molecules to mediate an adaptive response. Nevertheless, what determines the specificity to reacting with certain molecules is not yet known. Furthermore, underlying this question is the challenge in deciphering whether reactions of Fe-S clusters with signaling molecules are physiologically relevant or simply adventitious (e.g. as sometimes is the case in in vitro reactions). Although studies have definitively shown that some Fe-S regulatory proteins have specificity for more than one signal, here we focus mainly on the primary signaling molecule that enables these proteins to alter gene expression for ultimately achieving an adaptive response in vivo.

Figure 1.

Crystal structures of [2Fe-2S]²⁺-SoxR and apo-IscR bound to their respective DNA elements. In each structure, the protein dimer-DNA complex is shown as a cartoon representation, with monomeric protein subunits shown in cyan and green; DNA shown in gray; Fe-S cluster ligands highlighted in red; and, in the case of SoxR, the [2Fe-2S]²⁺ cluster shown as yellow spheres. (A) The active form of SoxR containing a solvent-exposed [2Fe-2S]²⁺ cluster. SoxR sharply bends the DNA, providing a mechanism for allowing RNAP to bind its promoter elements to initiate transcription. In contrast, while the inactive form of SoxR ([2Fe-2S]¹⁺-SoxR or apo-SoxR) can bind DNA, it presumably is unable to distort the DNA conformation. (B) The position of the cluster ligands within the apo-IscR structure suggests that in the holo-form of IscR, the [2Fe-2S] cluster would be solvent-exposed. Unlike SoxR, DNA bending is not thought to be part of the mechanism of transcriptional regulation by apo-IscR. This figure was prepared with PyMOL using structures available in the Research Collaboratory for Structural Bioinformatics Protein Data Bank under the accession codes 2ZHG ([2Fe-2S]²⁺-SoxR, ref [16]) and 4HF1 (apo-IscR, ref [97]).

Regulatory proteins modulate gene expression by a variety of mechanisms. For the purpose of this review, we will briefly describe mechanisms that are known or are hypothesized to be carried out by Fe-S regulatory proteins according to the bacterial paradigm. Commonly, genes are regulated at the transcriptional level by binding of regulatory proteins to specific DNA sequences in the promoter regions of genes. In doing so, these proteins (also referred to as transcription factors) can induce activation of gene transcription by recruiting RNAP to bind its target sites within the promoter either through protein-protein interactions or by altering the DNA architecture. On the other hand, transcription factor binding may occlude RNAP from recognizing its target promoter elements or, if RNAP-binding occurs simultaneously, interfere with subsequent events in initiation, thereby repressing transcription. In bacteria, the activity of RNAP can also be directly altered by different condition-specific sigma (σ) factors that associate with the RNAP core to form the RNAP holoenzyme active for site-specific DNA binding. Often times, these σ factors are inhibited by proteins called anti-sigma factors. Additionally, regulatory proteins can act in a relay to affect gene expression indirectly by altering activity of a downstream regulator. Such is the case for two component systems where in response to an environmental cue, a sensory protein modifies the activity of a cognate response regulator, which in turn directly alters transcription. Finally, gene expression may also be regulated at the post-transcriptional level. In this scenario, regulatory proteins or small RNAs bind to mRNA transcripts to either block translation or promote their stability, thus enabling translation to occur.

In recent years, much progress has been made in understanding the molecular mechanisms by which Fe-S clusters enable regulatory proteins to reprogram the expression of genes in response to environmental stimuli. Insightful reviews have focused on the chemistry by which Fe-S clusters of bacterial regulatory proteins specifically react with signaling molecules [5, 6, 19]. In this review, we briefly describe the major advancements achieved in exploring the cellular function of Fe-S regulatory proteins from both prokaryotes and eukaryotes, organized according to the primary signal to which they respond (see also Table 1).

Table 1.

Fe-S proteins in the regulation of gene expression.

Regulator	Organism	Cluster type	Primary Signal	Function
FNR	Proteobacteria and Bacilli	[4Fe-4S]	O2	Controls genes in the adaptive response to anaerobiosis
NreB	Staphylococci	[4Fe-4S]	O2	Phosphorylates the response regulator NreC to control genes in nitrate/nitrite respiration
AirS	Staphylococcus aureus strain Newman	[2Fe-2S]	O2	Alters anaerobic expression of genes involved in quorum sensing, virulence, and oxidative stress
SoxR	Proteobacteria and Actinobacteria	[2Fe-2S]	Redox-cycling compounds	Regulates the oxidative stress response directly or through the transcription factor SoxS
IscR	Proteobacteria	[2Fe-2S]	Fe-S cluster levels	Controls genes in Fe-S cluster biogenesis
SufR	Cyanobacteria	[4Fe-4S]	Fe-S cluster levels	Controls the sufBCDS Fe-S cluster biogenesis pathway
RirA	Rhizobia	Not yet determined	Fe levels	Regulates genes in Fe uptake
Fra2- Grx3/4	Saccharomyces cerevisiae	[2Fe-2S]	Fe levels	Controls activity of Fe-uptake regulators Aft1/2
Aft1/2	Saccharomyces cerevisiae	[2Fe-2S]	Fe levels	Regulates genes in Fe uptake
IRP1	Mammals	[4Fe-4S]	Fe levels	Alters stability or translation of transcripts involved in Fe uptake
Bacterial aconitases	Bacteria	[4Fe-4S]	Fe levels and/or ROS/NO	Alters stability or translation of transcripts, including those involved in Fe homeostasis, the oxidative stress, response, motility, or sporulation
NsrR	Proteobacteria, Bacilli, Streptomyces	[2Fe-2S] or [4Fe-4S]	NO	Regulates genes in the NO stress response
Wbl proteins	Actinobacteria	[4Fe-4S]	NO	Regulate genes involved in diverse cellular processes, including development, virulence, or antibiotic resistance
ArnR	Corynebacterium glutamicum	Not yet determined	NO	Controls genes in nitrate metabolism
RsmA	Streptomyces coelicolor	[2Fe-2S]	Not known	Inhibits activity of σM
ThnY	Sphingomonas macrogolitabida	[2Fe-2S]	Not known	Promotes activity of the tetralin utilization regulator ThnR
VnfA	Azotobacter vinelandii	[3Fe-4S]	Not known	Transcriptional activator of nitrogenase-2

Recent Achievements

Sensors of O₂

The availability of O₂ has a major impact on the regulation of several biological processes as O₂ serves as the terminal electron acceptor in aerobic respiration and can inadvertently give rise to toxic ROS. For many bacteria, distinct metabolic pathways are employed depending on the presence of O₂. While O₂ is the preferred terminal electron acceptor, when absent, fermentation or alternate terminal electron acceptors in anaerobic respiration are used to generate cellular ATP. Thus, the ability of bacteria to detect O₂ and reprogram the expression of aerobic, fermentative, or anaerobic respiratory systems is vital for survival in environments of varying O₂ tension. We describe three bacterial regulatory proteins that sense O₂ via their Fe-S cluster cofactor and respond accordingly through remodeling the transcriptome.

FNR

One of the best studied Fe-S regulatory proteins, the transcription factor FNR mediates an adaptive response when O₂ becomes limiting. In Escherichia coli, FNR controls the expression of ≥200 genes, activating those involved in anaerobic oxidation of carbon sources and reduction of alternate terminal electron acceptors (e.g., nitrate, fumarate, DMSO) and repressing genes specifically used in aerobic respiration [20–25]. FNR homologs are widely distributed in Proteobacteria and Bacilli and comparative genomic analyses indicate that the core FNR regulon is conserved across many facultative anaerobes [26, 27]. However, the variations that do exist even in closely related species may reflect adaptations to specific environments.

Under anaerobic conditions, E. coli FNR contains a [4Fe-4S]²⁺ cluster, which promotes FNR dimerization and subsequent binding to target promoters for transcriptional regulation. However, exposure to O₂ converts the [4Fe-4S]²⁺ to a [2Fe-2S]²⁺ cluster, thereby inactivating FNR through loss of dimerization. Thus, FNR directly senses O₂ through the lability of its [4Fe-4S]²⁺ cluster [28–32]. This O₂-dependent cluster conversion for FNR, in addition to that for O₂-sensitive [4Fe-4S] cluster-containing hydro-lyase enzymes [33], has established the paradigm for O₂ sensitivity of Fe-S clusters in general. Understanding the sensing mechanism is of widespread interest and has been intensively studied [19, 28, 34, 35]. Recent spectroscopic studies indicate that O₂ reacts with the [4Fe-4S]²⁺ cluster to form a [2Fe-2S]²⁺ cluster via sulfur-based oxidation [36]. Additional work has shown that the FNR [2Fe-2S]²⁺ cluster is sensitive to O₂⁻, which may be significant under aerobic conditions in preventing reconversion of [2Fe-2S]²⁺ back to [4Fe-4S]²⁺ and possibly explains why apo-FNR is the predominant form isolated from aerobic cells [37]. Furthermore, apo-FNR is specifically targeted for proteolysis, whereas [4Fe-4S]²⁺-FNR negatively regulates fnr transcription, suggesting that the pool of FNR protein is maintained at a level optimized for efficiently responding to fluctuations in O₂ [38, 39]. The effect of O₂ availability on FNR activity has also been computationally modeled, providing further insight into the design principles of the FNR regulatory network [40–45]. Finally, it appears that nucleoid-associated proteins (e.g., H-NS, IHF, Fis) and additional condition-specific transcription factors strongly influence FNR-dependent regulation on a genome-wide scale, alluding to the complexity of responding to O₂ in E. coli’s ever changing environment such as the mammalian gut [21].

The mechanism by which Gram-negative E. coli FNR senses O₂ is likely conserved since the majority of FNR homologs contain the four cysteine cluster ligands, three of which are near the N-terminus [46]. However, a few outliers exist, such as the case for FNR from Gram-positive Bacillus subtilis and Bacillus cereus. In contrast to EcFNR, BsFNR coordinates a [4Fe-4S] cluster by three C-terminal cysteines and one aspartate [47, 48]. While the O₂-sensitive cluster is required for BsFNR and BcFNR DNA binding, the proteins are dimerized even in their apo-protein forms [47–49]. Furthermore, apo- and holo-BcFNR bind to a subset of promoters with similar affinity in vitro, although this remains to be tested in vivo [49, 50]. Taken together, regulation of FNR activity in Bacilli substantially varies from that of E. coli FNR and other homologs and additional work is needed to understand the physiological significance behind this difference.

It should be noted that in addition to its primary function in mediating an adaptive response to O₂-limitation, FNR plays a role in sensing and responding to NO. In vitro, NO damages the E. coli FNR [4Fe-4S] cluster, resulting in decreased FNR DNA binding activity [51, 52]. In addition, cells exposed to NO exhibit defects in transcriptional regulation of a subset of FNR-dependent genes, including hmp, encoding the well-characterized flavohemoglobin involved in NO detoxification [51–54]. However, compared to the FNR response to O₂, the effect of NO on FNR activity appears lower in both amplitude and the number of FNR-regulated genes affected, suggesting that NO-sensing is a secondary function for FNR [51, 53, 54]. Rather, NorR and NsrR (discussed later in this review) are considered the primary transcription factors dedicated to sensing NO and controlling the expression of genes involved in NO detoxification and damage repair.

NreB

Instead of FNR, which is not present in Staphylococci, these Gram-positive bacteria employ the NreBC two-component system to activate transcription of genes involved in nitrate/nitrite reduction (narGHJI/nirRBD) and transport (narT)[55, 56]. Staphylococcus carnosus NreB is a histidine kinase that upon autophosphorylation, transfers the phosphoryl group to the response regulator NreC, which subsequently modulates gene expression through DNA-binding [55]. NreB activity depends on the presence of a [4Fe-4S] cluster coordinated by four cysteines [57, 58]. Upon O₂-exposure, the [4Fe-4S] cluster converts to a [2Fe-2S] cluster that is, in turn, degraded to generate inactive, apo-NreB [58]. Thus, like FNR, NreB directly senses O₂ through its Fe-S cluster. However, while FNR directly controls transcription, NreB indirectly alters gene expression through NreC phosphorylation. It was recently demonstrated that NreB activity is also altered by the nitrate receptor NreA. Unlike nitrate-bound NreA, NreA in the nitrate-free form directly interacts with NreB to inhibit autophosphorylation [59]. Thus, these O₂- and nitrate-dependent modes of NreB regulation enable Staphylococci to carry out nitrate respiration under appropriate growth conditions.

AirS

AirSR of Staphylococcus aureus strain Newman was recently proposed to be an O₂-sensing two-component system. However, in contrast to NreBC that regulates a limited gene set, transcriptomic analysis indicates that AirSR is a global regulatory system, altering anaerobic expression of numerous genes with diverse functions, such as quorum sensing, virulence, oxidative stress [56, 60]. In vitro studies indicate that the histidine kinase AirS contains an O₂-sensitive [2Fe-2S] cluster that is required for AirS autophosphorylation and subsequent transfer of the phosphoryl group to AirR [60]. Curiously, a separate study found that AirSR of S. aureus NCTC8325 influences vancomycin resistance and controls expression of cell wall synthesis genes primarily under aerobic growth conditions [61]. Furthermore, research from a third group suggests that AirSR (referred to as YhcSR) is essential in S. aureus WCUH29 and regulates distinct genes (lac operon) or those in common with AirSR of S. aureus NCTC8325 (nreABC and narGHJI) and AirSR of S. aureus Newman (opuCAB)[60–63]. Thus, additional work is needed to determine if this difference in AirSR function can be attributed to S. aureus strain differences.

Perspective on the scope of O₂-sensing

Outside of the laboratory setting, bacteria are routinely exposed to environments with varying O₂ tension. Thus, establishing the identity of O₂-regulated genes and the mechanism by which they are controlled will be key to understanding how individual bacteria survive in nature. While the O₂-sensing proteins reviewed here have set the framework for determining how a relatively small set of bacteria accomplish this task, it is highly likely that future studies will reveal the existence of many more regulatory proteins that use Fe-S clusters to sense O₂.

SoxR, a sensor of redox-cycling compounds

Widely distributed in Proteobacteria and Actinobacteria, the well-known transcription factor SoxR utilizes its [2Fe-2S] cluster to sense and respond to oxidative stress. Although apo- and holo-forms can bind DNA, SoxR is activated by one electron oxidation of its [2Fe-2S] cluster to the +2 state and subsequently reorients the promoter DNA elements to allow RNAP to initiate transcription [64, 65]. Early studies suggested that O₂⁻ was responsible for cluster oxidation; however, recent work demonstrated that redox-cycling compounds, naturally produced by some bacteria, fungi, and plants, are the more likely inducer of SoxR activity [66]. In enteric bacteria, SoxR mediates a general stress response by activating expression of the global regulator SoxS that controls transcription of ≥100 genes, including those encoding Mn²⁺-superoxide dismutase, DNA repair nucleases, oxidation-resistant enzymes, and efflux pumps [64, 65]. However, nonenterics lack SoxS and SoxR takes on the direct role of regulating a relatively small set of genes (encoding transporters, NADPH-dependent reductases and monooxygenases) in response to endogenously produced redox-cycling compounds, such as pyocyanin by Gram-negative Pseudomonas aeruginosa and actinorhodin by Gram-positive Streptomyces coelicolor [67–70]. Interestingly, SoxR homologs exhibit different sensitivities to various redox-cycling agents. For example, E. coli SoxR is activated by a wide-range of redox-cycling compounds, consistent with the crystal structure indicating that the [2Fe-2S] cluster is solvent-exposed (Fig. 1A) [16, 71, 72]. In contrast, the [2Fe-2S] cluster of S. coelicolor SoxR exhibits a higher redox potential, and accordingly, is more limited in the range of chemicals to which it responds [71, 72]. In addition, despite the clusters of E. coli and P. aeruginosa SoxR proteins having similar redox potentials, PaSoxR expressed in E. coli displayed reduced sensitivity to paraquat and three key residues in PaSoxR that are not conserved in EcSoxR were found to contribute to this phenotype [64, 71]. However, a subsequent study demonstrated that although paraquat activated PaSoxR more slowly than EcSoxR, these two SoxR proteins behaved similarly in responding to paraquat, among a broad range of other compounds, when expressed in S. coelicolor or E. coli [72]. Thus, further work is needed to address what additional factors, including residues or structural differences among SoxR homologs, may give rise to alterations in SoxR behavior and redox potential. Although the physiological significance is not yet known, it has been demonstrated that activation of SoxR can also occur by a DNA-mediated charge transfer reaction, in which guanine radicals serve as the oxidant, and by exposure to NO [73, 74].

Sensors of Fe or Fe-S clusters

For most organisms, Fe is indispensable as it serves as a cofactor, in the forms of Fe ions, heme, or Fe-S clusters, for proteins vital to various biological processes. However, excess amounts of Fe are harmful due to its ability to catalyze the formation of ROS that damage cellular components. Thus, much research has focused on dissecting the regulatory systems required for maintaining homeostasis of Fe-containing cofactors. Below, we list several examples of Fe-S regulatory proteins that modulate gene expression in response to alterations in the cellular concentration of Fe or Fe-S clusters.

IscR

In E. coli, IscR uses the ligation state of its own [2Fe-2S] cluster to sense the cellular Fe-S status and adjust transcription of relevant genes accordingly. IscR was initially discovered as a repressor of the iscRSUAhscBAfdx (isc) operon, encoding IscR and the housekeeping Isc Fe-S biogenesis pathway, and the homeostatic mechanism by which IscR regulates this operon has been investigated in detail [75, 76]. Upon acquiring a [2Fe-2S] cluster from the Isc machinery, holo-IscR binds the iscR promoter to repress transcription. As some Fe-S clusters are sensitive to O₂ and/or ROS, the presence of these oxidants increases the level of substrate proteins that need Fe-S biogenesis or repair. As a result, conditions that increase the Fe-S demand are predicted to elevate the competition between IscR and other apo-protein substrates for the Isc proteins. This would lead to a population of IscR protein that is less occupied with [2Fe-2S] clusters, resulting in de-repression of isc. Thus, expression of the Isc pathway is coupled to the Fe-S demand through the levels of the IscR [2Fe-2S] cluster. While current evidence strongly supports this homeostatic model, it does not exclude the possibility that stability of the IscR [2Fe-2S] cluster may also be directly affected by ROS or Fe-limitation, factors both known to promote isc expression [77–82].

IscR also controls expression of several other Fe-S assembly factors, Fe-S enzymes, the Mn²⁺-containing enzymes superoxide dismutase and ribonucleotide reductase, and additional genes with diverse cellular functions, such as biofilm formation, colicin K production, and RNA metabolism [83–87]. Furthermore, there is growing evidence that IscR homologs in several pathogenic bacteria play a role in virulence [88–92]. An unexpected property of IscR is that a subset of IscR-controlled genes is regulated by either holo- or apo-IscR, unlike isc [93]. For example, in vivo and in vitro evidence demonstrates that the [2Fe-2S] cluster is not required for IscR to activate sufABCDSE, encoding the alternate, stress-induced Suf Fe-S biogenesis pathway [78, 93]. This differential regulation is at least in part due to the ability of IscR to recognize two distinct DNA motifs, referred to as Type 1 and Type 2 sites [83]. While promoters with Type 1 sites (e.g., iscR) exhibit a strict requirement for holo-IscR, promoters with Type 2 sites (e.g., sufA) are bound by apo- and holo-forms of IscR with equally high affinity [76, 93]. Thus, IscR is a novel type of regulator in that multiple forms of the protein are active to differentially regulate transcription through binding of two distinct DNA motifs. While work is ongoing to understand the Fe-S cluster-induced structural changes that may allow IscR to mediate differential gene expression (discussed below), the physiological significance of this mechanism is that IscR regulates transcription under all growth conditions but switches the genes it regulates. Such a mechanism potentially allows E. coli to maintain Fe-S homeostasis under any stress condition that increases the Fe-S demand.

Prevalent in Proteobacteria, IscR is a member of the Rrf2 family of transcription factors [94]. While this family has not been extensively studied, sequence comparisons have revealed that a subset contain three conserved cysteine residues in their C-terminal region [95]. In fact, these three cysteines plus a histidine serve as ligands for the IscR [2Fe-2S] cluster [93, 96]. Several crystal structures of Rrf2 proteins have been solved, including that of apo-IscR (Fig. 1B) [95, 97–99]. Although the electron density of residues encompassing the IscR cluster-binding region was weak, its location indicated that the cluster-binding site is solvent-exposed, perhaps allowing for efficient incorporation and/or loss of the [2Fe-2S] cluster. Additionally, the cluster-binding site from one monomer is proximal to the DNA binding domain of the other monomer, raising the possibility that cluster ligation induces a conformational change within IscR that influences DNA binding. This may involve rearrangement of glutamate 43 in the IscR DNA binding domain that was found to be a critical residue in discriminating between Type 1 and Type 2 sites [97]. While these findings have provided the initial framework to understanding how IscR distinguishes between two DNA motifs, more work is needed to address if this trait is exhibited by other Rrf2 proteins that potentially ligate Fe-S clusters, as discussed later in this review.

SufR

In Gram-negative Cyanobacteria, the [4Fe-4S] cluster-containing SufR protein negatively regulates expression of the sufBCDS operon, encoding the primary Fe-S biogenesis pathway of these bacteria, and its own gene, sufR [100–102]. In vitro DNA binding by SufR appears to depend on both the presence and oxidation state of the cluster; [4Fe-4S]²⁺-SufR bound with high affinity to the sufB and sufR promoter regions, unlike [4Fe-4S]⁺¹-SufR or apo-SufR [100]. Although SufR and E. coli IscR belong to different protein families (the DeoR family of helix-loop-helix proteins and the Rrf2 family, respectively), SufR is proposed to regulate sufBCDS expression through a homeostatic mechanism similar to that of IscR in regulating the iscRSUAhscBAfdx operon [101]. Further work is needed to address this and whether SufR controls the transcription of additional genes in Cyanobacteria. Nevertheless, it is an interesting parallel that the housekeeping Fe-S biogenesis pathways of E. coli and Cyanobacteria are both subject to transcriptional repression by a Fe-S dependent transcription factor.

RirA

While most bacteria maintain Fe homeostasis primarily through the regulatory action of the Fe²⁺-sensing protein Fur, some members of the Gram-negative Rhizobia utilize the transcription factors Irr and RirA [94, 103, 104]. In Rhizobium leguminosarum, Sinorhizobium meliloti, and Agrobacterium tumefaciens, RirA has a global role in regulating the expression of genes involved in iron/heme uptake and siderophore synthesis [105–111]. Although the presence of a Fe-S cluster within RirA has yet to be confirmed, RirA, like IscR, is a Rrf2 family member and contains the three conserved C-terminal cysteine residues [95]. Indeed, a recent study demonstrated that these cysteines are essential for RirA-mediated repression of the sufS2BCDS1XA operon, encoding the Fe-S cluster biogenesis pathway of A. tumefaciens [112]. Furthermore, sufS2, encoding a putative cysteine desulfurase, is also required for RirA to repress this pathway, suggesting that in addition to maintaining Fe homeostasis, RirA may function similarly to IscR in maintaining Fe-S cluster homeostasis. In contrast, sufS2 was dispensable for RirA-dependent regulation of Fe acquisition and siderophore synthesis genes, thus raising the question as to the mechanism by which RirA differentially controls gene expression [112]. While further work is needed to address this question and to determine the RirA Fe-S cluster type, RirA-mediated gene regulation is clearly reminiscent of E. coli IscR. Curiously, while most α-Proteobacteria contain IscR homologs, they do not appear to contain both iscR and rirA, thus raising the possibility that RirA may have evolved to control a more diverse gene set in some Rhizobia [94]. Indeed, the fourth His107 cluster ligand and most residues important for specific DNA-interactions in E. coli IscR are not conserved in RirA (Fig. 2)[96, 97].

Figure 2.

Comparison of protein sequences for E. coli IscR (GenBank accession number NP417026) and RirA from S. meliloti (AAF06014); R. leguminosarum (CAC35510); and A. tumefaciens (AAK86020) created using Clustal Omega[187–189]. In red are residues that are known or are predicted to ligate a Fe-S cluster. In blue are conserved residues that are important for IscR to make direct interactions with DNA.

Fra2-Grx3/4 and Aft1/2

In the budding yeast Saccharomyces cerevisiae, Fe homeostasis is maintained by the paralogous transcription factors Aft1 and Aft2, which have overlapping and distinct DNA targets. During Fe-limitation, Aft1 (and presumably Aft2) accumulates in the nucleus to activate transcription of Fe uptake and storage genes. In contrast, when Fe levels are sufficient, multimerization of Aft1/2 promotes their export from the nucleus, thereby inhibiting Aft1/2-mediated regulation. In recent years, much work has shed light on the mechanism by which Aft1/2 activities are controlled, which was previously known to rely on mitochondrial Fe-S biogenesis and involve the cytosolic monothiol glutaredoxins Grx3 and Grx4, and a homolog of the E. coli BolA protein, Fra2 [113, 114].

Either Grx3 or Grx4 interact with Fra2 to form a heterodimer with a bridging [2Fe-2S] cluster, coordinated by glutathione, a Grx3/4 cysteine, and a Fra2 histidine [115, 116]. These residues, and those of the Grx3/4 glutathione-binding pocket, are also essential for Fe-dependent inhibition of Aft1 activity in vivo [117]. The possible mechanism behind this observation was recently made clear by in vitro experiments demonstrating that Aft2 specifically interacts with Fra2-Grx3 and that the [2Fe-2S] cluster of the heterodimer is transferred to Aft2. Upon cluster acquisition, Aft2 dimerizes, which in turn decreases its DNA-binding affinity [118]. Thus, under Fe-limiting conditions in which levels of Fe-S clusters may be insufficient, monomeric apo-Aft2 is predicted to be available to upregulate expression of Fe uptake genes. Together, these findings have been critical in directly linking S. cerevisiae Fe-S cluster biogenesis to the cellular Fe status.

Interestingly, there is growing evidence that the relationship between monothiol glutaredoxins and BolA-like proteins is widely conserved among eukaryotes and prokaryotes [113, 119–121]. In fact, in addition to S. cerevisiae, [Fe-S]-bridged heterodimers form between the respective Grx and BolA-like proteins from human, E. coli, and Arabidopsis thaliana [122–124]. Further work is needed to define the cellular roles of these heterodimers and establish if the Fe-S cluster-dependent mechanism of regulation is conserved in other organisms.

IRP1 and bacterial aconitases

Another example connecting Fe-S cluster biogenesis with Fe homeostasis is that of the well-studied mammalian protein IRP1, extensively reviewed in [125–128]. In response to Fe deficiency, IRP1, along with IRP2, increase cellular Fe levels through post-transcriptional regulation of genes involved in Fe uptake and storage, among numerous other functions. IRP1/2 bind to stem-loop structures called iron-responsive elements (IRE) in the 3′ or 5′ untranslated regions of mRNA transcripts, thereby altering transcript stability or translation. In the case of IRP1, this activity is modulated through a [4Fe-4S] cluster. Under Fe-replete conditions, IRP1 contains a [4Fe-4S] cluster and functions as a cytosolic aconitase that does not bind RNA. The cluster is coordinated by only three cysteines, such that one labile Fe is available for the citrate-isocitrate isomerization carried out by aconitase. When cellular Fe levels decrease, the cluster may be susceptible to losing this Fe, generating the inactive [3Fe-4S]-aconitase. Subsequently, the [3Fe-4S] cluster can be degraded, resulting in an apo-protein form (IRP1) that binds IREs with high affinity. Disruption of Fe-S cluster assembly pathways promotes IRP1 IRE binding activity, as does exposure to O₂, ROS, and RNS. In addition, IRP1 activity is altered through phosphorylation and protein stability. Thus, IPR1 is highly controlled through multiple mechanisms to ensure optimal maintenance of Fe homeostasis. It is intriguing that this dual aconitase/regulator function is also present in some bacteria, recently reviewed in [6].

Sensors of NO

Nitric oxide (NO) is a membrane-permeable, free radical gas. In bacteria, NO is a natural byproduct released during the respiratory denitrification pathway. Additionally, NO serves as an important signaling molecule in both prokaryotes and eukaryotes. However, at high concentrations, NO can be toxic as its high chemical reactivity damages DNA and amino acids. Furthermore, NO destroys Fe-S clusters and reacts with Fe to form Fe-nitrosyl compounds. Indeed, phagocytes synthesize NO as a defense molecule against pathogenic bacteria. Because of the deleterious effects caused by NO, organisms have evolved ways to combat nitrosative stress [129–131]. Below we describe three examples of bacterial transcription factors that sense NO through their Fe-S cluster and mediate adaptive responses by controlling the expression of genes involved in NO detoxification and repair, or various other cellular processes.

NsrR

Perhaps the most studied NO-sensing regulator, NsrR is broadly distributed in most β- and γ-Proteobacteria, in addition to some Gram-positive bacteria, such as Bacillus and Streptomyces species [132]. In addition, like IscR and RirA, NsrR belongs to the Rrf2 family of transcription factors [95]. Determination of the NsrR regulon in several bacteria has revealed that NsrR commonly controls the expression of genes involved in NO detoxification and damage repair, including hmp, encoding a flavohemoglobin; ytfE, implicated in Fe-S cluster repair; and, nrf, encoding the NrfA periplasmic nitrite reductase [132–140]. NsrR also regulates genes important for diverse cellular processes, including motility and biofilm development; glyoxal resistance; virulence; and bacteria-host symbiosis [135, 139, 141–143].

Isolated NsrR proteins of S. coelicolor, B. subtilis, and Neisseria gonorrhoeae all contain an NO-sensitive Fe-S cluster that is required for site-specific DNA binding [144–147]. Furthermore, mutation of at least one of the three conserved C-terminal cysteine residues results in loss of the cluster, DNA binding, and/or NsrR-mediated gene regulation [135, 146, 147]. Despite these common features, the type of the NsrR cluster remains in question; while isolated NsrR of S. coelicolor and N. gonorrhoeae contains a [2Fe-2S] cluster, B. subtilis NsrR contains a [4Fe-4S] cluster [144–147]. Variations in purification methods may account for this difference or, as the spacing of the conserved cysteines slightly varies, it is possible that the cluster type may differ among NsrR orthologs [133].

Interestingly, B. subtilis NsrR binds to a subset of promoters in a NO-insensitive manner in vitro, raising the possibility that apo-NsrR may also be functional [134]. However, Henares et al later showed that in vivo, these promoters are either bound by NsrR in a NO-sensitive manner or are not true NsrR targets [136]. While further work is required to resolve this issue, it is nevertheless tempting to compare NsrR with IscR, in which both apo- and holo-forms are functional (see above). Furthermore, like IscR, NsrR from E. coli and B. subtilis binds two different classes of DNA target sites. However, while IscR binds two motifs with distinct sequences, NsrR binds two sites with similar sequences: one containing an inverted repeat, and a second containing one half site of the inverted repeat [135, 136]. Work is ongoing to determine whether NsrR has different binding affinities for these sites, to identify the fourth non-cysteinyl cluster ligand, and to resolve the mechanism by which the NsrR cluster senses NO.

Wbl proteins

Initially characterized in S. coelicolor, homologs of WhiB (named Wbl for WhiB-like proteins) are highly conserved and exclusive to Actinobacteria and growing evidence demonstrates that this protein family exhibits functional roles in numerous processes such as cell division, sporulation, nutrient starvation, antibiotic resistance, virulence, and oxidative stress responses [6, 148–163]. Despite differences in their expression, structure and function, Wbl proteins commonly ligate a Fe-S cluster through four conserved cysteine residues [164, 165]. S. coelicolor WhiD and the WhiB1, WhiB3, and WhiB4 proteins of Mycobacterium tuberculosis contain a NO-sensitive [4Fe-4S] cluster and recent studies have provided insight on the mechanism of cluster nitrosylation in which reaction of NO with the cluster of Wbl proteins forms a pair of dinuclear iron tetra-nitrosyl species known as Roussin’s red ester complexes [19, 158, 166–170]. Following cluster nitrosylation or when in their apo-protein forms, Wbl proteins bind DNA with high affinity [153, 158, 169, 171]. Furthermore, for WhiB3, WhiB4, and WhiB1, it has been demonstrated that in the absence of the cluster, formation of two intramolecular disulfide bonds is critical for high affinity DNA binding [153, 158, 169]. Thus, a current regulatory model for these proteins proposes that NO-mediated damage of the Fe-S cluster converts the inactive, holo-protein form to an oxidized, apo-protein capable of binding DNA to repress (WhiB1, WhiB4) or activate (WhiB3) transcription [153, 158, 169, 172]. In contrast, exposure of holo-WhiB7 to the oxidant diamide decreased WhiB7-dependent activation, suggesting that loss of the Fe-S cluster results in an inactive form of WhiB7 [173]. Differences in in vitro DNA-binding specificity also appear to exist among Wbl proteins: while WhiB1 and WhiB2 were shown to bind DNA in a site-specific manner, WhiB3 and WhiB4 can non-specifically bind DNA [153, 158, 169, 171, 172]. Thus, the specific regulatory mechanisms used to modulate gene expression may be unique for each Wbl protein, thereby underlying the diverse cellular roles exhibited by this class of proteins. Additionally, some Wbl proteins display additional functions, such as disulfide reductase activity and chaperone-like activity [164, 174, 175]. The multifunctionality of these proteins, the physiological significance of sensing NO, and identification of direct Wbl regulons require further study.

ArnR

Recently, ArnR of Gram-positive Corynebacterium glutamicum has been identified as a novel transcription factor for the regulation of anaerobic nitrate reduction. Similar to the Wbl proteins, ArnR orthologs are conserved in Actinobacteria, primarily other Corynebacterium, Mycobacterium, and Nocardioides species [176]. Furthermore, ArnR shares no significant homology with known NO-responsive regulators, including NsrR, which is not present in C. glutamicum [132, 176]. Nevertheless, ArnR represses hmp and narKGHJI and isolated ArnR exhibits an absorption spectrum characteristic of a Fe-S protein [176]. Exposure to NO or mutation of three putative cysteine cluster ligands disrupts this absorption spectrum, in addition to ArnR DNA binding in vitro and ArnR-dependent gene regulation in vivo [177]. Thus, ArnR activity may be regulated by NO through its putative Fe-S cluster. Further characterization of the ArnR cluster and its role in NO-sensing are needed to gain additional insight as to how C. glutamicum may cope with NO stress.

Additional sensors

Below we briefly discuss three bacterial Fe-S regulatory proteins whose in vivo signaling molecule remains to be established:

RsmA

RsmA of S. coelicolor is the first example of an anti-sigma factor whose activity is modulated through its [2Fe-2S] cluster. In its holo-form, RsmA interacts with the alternative sigma factor σ^M to prevent its association with RNAP, and thus, σ^M-mediated transcription [178]. Compared to holo-RsmA, apo-RsmA exhibits distinct structural differences, suggesting that cluster loss may cause an altered protein conformation with decreased affinity for σ^M [178]. Although the cellular role for σ^M has yet to be defined, its expression is induced by osmotic stress, raising the possibility that the RsmA [2Fe-2S] cluster may sense a signal produced under these growth conditions [179].

ThnY

In response to the organic solvent tetralin (1,2,3,4-tetrahydronaphthalene), the [2Fe-2S] cluster-containing flavoprotein ThnY and the transcriptional activator ThnR promote expression of the thn genes required for tetralin utilization in Gram-negative Sphingomonas macrogolitabida [180, 181]. In vitro, the presence of ThnY increased ThnR binding affinity for promoter DNA, suggesting that ThnY may indirectly activate thn transcription through protein-protein interactions with ThnR [181]. While the significance of the ThnY [2Fe-2S] cluster in this mechanism is not known, it has been proposed that ThnY activity could be dictated by the cluster redox state, possibly regulated by the ferredoxin ThnA3. Indeed, mutation of Cys40 in the putative cluster-binding domain resulted in a ThnY variant capable of activating thn expression in response to compounds other that tetralin, implying a role of the [2Fe-2S] cluster in proper regulation of this pathway [182].

VnfA

VnfA of Gram-negative Azotobacter vinelandii is a transcriptional activator required for expression of the alternative nitrogenase-2 enzyme. VnfA consists of three domains: an N-terminal GAF domain also containing a cysteine-rich motif that ligates a [3Fe-4S] cluster; a central AAA+ NTPase domain; and a C-terminal DNA binding domain [183]. While both apo- and holo-forms bind target DNA, holo-VnfA exhibits increased ATPase activity that, in turn, is required for VnfA to activate transcription. In contrast, in the absence of the [3Fe-4S] cluster, the GAF domain inhibits ATPase activity and thus VnfA-mediated regulation [184]. Therefore, the presence of the cluster is proposed to block inhibition by the GAF domain. While the signal to which the [3Fe-4S] cluster responds is not yet known, a decrease in VnfA transcription activation was observed in cells treated with the O₂⁻ generator PMS, suggesting that ROS may play a role in this unique regulatory mechanism [183]. As far as we are aware, VnfA is the first example of a transcription factor whose function is controlled by a [3Fe-4S] cluster.

Outlook

Clearly, Fe-S regulatory proteins play prominent roles in the ability of organisms to efficiently adapt to their environment. Although many research breakthroughs have shed light on the mechanisms by which Fe-S clusters control regulatory protein function, there is much yet to be learned in this area of investigation. For example, there is in general a lack of structural information on how the ligation or oxidation state of the Fe-S cluster alters protein conformation. While the cluster-containing and apo-protein crystal structures of IRP1 have been solved, and while preliminary studies suggest conformational changes within SoxR occur based the oxidation state of its cluster, further structural studies are needed to explain how Fe-S clusters modulate regulatory protein activity [16, 17, 185, 186]. Along the same lines, how the cluster environment tunes the sensitivity of Fe-S clusters toward specific signaling molecules is not well understood. In addition, we are just now beginning to understand the regulatory mechanisms by which Fe-S clusters are assembled and transferred to apo-protein substrates. As these processes can occur by different pathways, this adds another layer of complexity as to how the function of Fe-S regulatory proteins is regulated.

Highlights.

• We review Fe-S cluster proteins that regulate gene expression.

• These proteins sense environmental signals through their Fe-S cluster.

• The status of the Fe-S cluster dictates activity of the regulatory protein.

• This mechanism enables cells to rapidly adapt to their environment.