Ferrous Iron Efflux Systems in Bacteria

Abstract

Bacteria require iron for growth, with only a few reported exceptions. In many environments, iron is a limiting nutrient for growth and high affinity uptake systems play a central role in iron homeostasis. However, iron can also be detrimental to cells when it is present in excess, particularly under aerobic conditions where its participation in Fenton chemistry generates highly reactive hydroxyl radicals. Recent results have revealed a critical role for iron efflux transporters in protecting bacteria from iron intoxication. Systems that efflux iron are widely distributed amongst bacteria and fall into several categories: P_1B-type ATPases, cation diffusion facilitator (CDF) proteins, major facilitator superfamily (MFS) proteins, and membrane bound ferritin-like proteins. Here, we review the emerging role of iron export in both iron homeostasis and as part of the adaptive response to oxidative stress.

Graphical abstract

Introduction

Iron is critical for cell growth and survival. However, when present in excess, it is also detrimental to cells. Under aerobic conditions, iron toxicity is closely related to oxidative stress through Fenton chemistry¹. Hydrogen peroxide (H₂O₂) reacts with ferrous iron (Fe²⁺) to generate highly reactive hydroxyl radicals that damage macromolecules such as DNA, proteins and fatty acids, resulting in disruption of cell metabolism and ultimately cell death². Therefore, the toxicity of reactive oxygen species (ROS) is generally thought to be exacerbated by conditions that elevate the intracellular iron pool. Conversely, high levels of intracellular iron may also be toxic independent of ROS, presumably due to the ability of iron to compete with other transition metals, such as manganese, for binding to metal-dependent enzymes or regulators, resulting in mismetallation and inactivation of these proteins^{3, 4}. ROS such as H₂O₂ and superoxide radical can disrupt iron-sulfur clusters and mononuclear iron centers of iron-enzymes, thereby leading to iron release^{5, 6}. Therefore, iron intoxication may also be exacerbated by an elevation in ROS. Clearly, the toxicity of iron and ROS are closely intertwined, with each potentially increasing the toxicity of the other.

Bacteria adapt to environmental stresses by activation of specific transcriptional programs. In the case of iron homeostasis, bacteria monitor intracellular iron levels using metal-sensing (metalloregulatory) proteins^{7, 8}. The ferric uptake regulator (Fur) protein is the most widespread bacterial iron sensor⁹, but it can be replaced by functionally analogous proteins such as IdeR (in actinomycetes)^{10, 11} and Irr (in alpha-proteobacteria)^12–14. Fur helps to maintain iron homeostasis by regulating genes implicated in iron uptake, storage, and efflux¹⁵. Typically, Fur is considered to function as an Fe²⁺-activated transcriptional repressor for most of its targets, but there are increasing examples where Fur functions as a transcriptional activator or where it binds DNA in the absence of bound iron^16–18.

Iron-sensing regulators such as Fur play a central role in the control of iron homeostasis¹⁹. The Escherichia coli Fur regulon illustrates the diverse roles that Fur may play. E. coli Fur (Fur_EC) binds to DNA when associated with Fe²⁺ and serves to repress the expression of target operons²⁰. This repression is relieved under iron-limited conditions, and this results in the derepression of iron uptake systems, including the synthesis of the high-affinity iron-chelating compound siderophore known as enterobactin and its cognate import system²¹. Fur also helps bacteria to remodel their proteomes to prioritize the utilization of iron, in a process known as "iron-sparing" (Fig. 1)^22–24. In E. coli, the loss of Fur_EC DNA-binding activity (under low iron conditions) results in expression of the RyhB small RNA (sRNA) that represses translation of non-essential iron-enzymes^22–24. Fur also participates in the regulation of gene expression under conditions of iron excess. For example, Fur_EC positively regulates expression of the iron storage protein ferritin by occluding the binding of the H-NS transcriptional repressor²⁵. In general, adaptation to iron excess often involves expression of iron storage functions (including heme-containing bacterioferritins, ferritins, and Dps-family mini-ferritins) but may additionally require iron efflux systems (Fig. 1). In light of the central role of Fur in coordinating iron homeostasis, it is not surprising that some iron efflux systems are induced by Fur in response to iron excess^{26, 27}.

fig. 1. Iron homeostasis in bacteria.

Under iron deficient conditions (left), high affinity iron uptake systems are induced to scavenge iron from the surroundings to maintain the cell's labile iron pool. when iron is limiting, it is selectively partitioned to the most essential functions and incorporation into lower priority iron enzymes is translationally inhibited as part of an iron sparing response. in many cases, iron-independent enzymes may be derepressed to replace functions that would otherwise depend on iron. under iron excess conditions, the cell will have a full complement of iron-requiring enzymes, and iron in excess of immediate needs will be either stored for future use or exported by fe²⁺ efflux transporters to prevent iron overload.

Bacteria also adapt to oxidative stress by the induction of specific defensive genes. For example, H₂O₂ induces a specific peroxide-stress response that is regulated by the OxyR repressor in E. coli²⁸ and by the PerR repressor in Bacillus subtilis²⁹. In both model organisms, a rise in intracellular H₂O₂ triggers the induction of defensive enzymes such as catalase and alkyl hydroperoxide reductase, which can directly detoxify H₂O₂. In addition, cells scavenge excess iron from the cytosol by sequestration into mini-ferritin proteins, including Dps in E. coli³⁰ and the Dps ortholog MrgA in B. subtilis³¹. The co-regulation of H₂O₂ degradation enzymes and iron-sequestering proteins further highlights the central role of iron in peroxide intoxication. In addition to scavenging iron, peroxide stress also frequently modulates metal uptake and efflux systems³². In E. coli, H₂O₂ induces an OxyR-activated Mn²⁺ uptake system (MntH)^{33, 34}, and in B. subtilis H₂O₂ induces a PerR-regulated iron efflux system, PfeT^{35, 36}. PfeT is a member of the P_1B4-type ATPases, and recent results indicate that several close homologs also function as Fe²⁺ efflux pumps^{27, 37–39}. Fe²⁺ efflux pumps have now been documented in a wide variety of bacteria, and include P_1B-type ATPases, cation diffusion facilitator (CDF) proteins, major facilitator superfamily (MFS) proteins, and membrane bound ferritin-like proteins (Table 1 & Fig. 2). Here, we summarize the emerging role of these ferrous iron efflux pumps in helping ameliorate the deleterious effects of excess iron and peroxide.

Table 1.

Fe²⁺ efflux transporters in bacteria.

Protein	Organism	Function	Category	Substrate specificity*	Transcription regulation	References
PfeT	Bacillus subtilis	Fe2+ efflux	P1B4-type ATPase	Fe2+ a, b, Co2+ a	PerR & Fur	36
FrvA	Listeria monocytogenes	Fe2+ efflux	P1B4-type ATPase	Fe2+ a, b, Co2+ a, Zn2+ a	PerR & Fur	27
CtpD	Mycobacterium tuberculosis	Fe2+ efflux	P1B4-type ATPase	Fe2+ a, b, Co2+ a	Unknown	37
PmtA	group A Streptococcus	Fe2+ efflux	P1B4-type ATPase	Fe2+ b	PerR	38, 39
Nia	Sinorhizobium meliloti	Fe2+ or Ni2+ efflux	P1B5-type ATPase	Fe2+ a, b, Ni2+ a, b	Unknown	73
FieF (YiiP)	Escherichia coli	Fe2+ or Zn2+ efflux	CDF family	Fe2+ b, Zn2+ a, Cd2+ a	Unknown	92–94
AitP	Pseudomonas aeruginosa	Fe2+ or Co2+ efflux	CDF family	Fe2+ b	Unknown	97
FeoE	Shewanella oneidensis	Fe2+ efflux	CDF family	Fe2+ b	Unknown	101
IceT	Salmonella typhimurium	Fe2+ citrate or citrate efflux	MFS family	Fe2+ b	BaeSR	120
MbfA	Agrobacterium tumefaciens	Fe2+ efflux	Membrane bound ferritin	Fe2+ b	Irr	125
MbfA	Bradyrhizobium japonicum	Fe2+ efflux	Membrane bound ferritin	Fe2+ b	Irr	127

^*Note: the substrate specificity of the transporters is either based on biochemical measurements (a), inferred from physiology studies (b), or both (a, b).

fig. 2. Ferrous iron efflux systems in bacteria.

Four different groups of transporters can function as fe²⁺ efflux pumps. i. p_1b-atpase; ii. cation diffusion facilitator (cdf); iii. major facilitator superfamily (mfs); iv. membrane-bound ferritin. a typical p_1b-atpase consists of a transmembrane domain (tmd) that has 6–8 helices, a soluble actuator domain (not shown), and an atp-binding domain (atp-bd)⁵². a cdf transporter contains a n-terminal domain (ntd), a transmembrane domain (tmd) that has 6 helices, a histine-rich interconnecting loop (il) between tm4 and tm5 (not shown), and a c-terminal cytoplasmic domain (ctd)⁷⁵. the common structural fold (mfs fold) of a mfs transporter is composed of two distinct domains, n domain and c domain. each domain has six consecutive transmembrane helices¹⁰³. a membrane-bound ferritin transporter has two major domains, n-terminal ferritin-like or er domain (er) and c-terminal membrane-embedded vacuolar iron transporter domain (vit1).

P-type ATPases

The P-type ATPases are a large group of transmembrane proteins that transport ions and lipids across cellular membranes, energetically driven by ATP hydrolysis⁴⁰. Five subgroups of P-type ATPases have been defined based on sequence homology and substrate specificity⁴¹. These are the P₁-type (K⁺ and transition metal transporters), P₂-type (Ca²⁺, Na⁺/K⁺, and H⁺/K⁺ pumps), P₃-type (H⁺ pumps), P₄-type (phospholipid transporters), and P₅-type ATPases (unknown substrate). The P₂-type ATPases have been well studied and are more prevalent in eukaryotes than in prokaryotes. The majority of P₃-type ATPases are H⁺ pumps found in plants and fungi. Some of the P₄-type ATPases have been revealed to be phospholipid transporters^{42, 43}. No specific substrate has yet been identified for the P₅-type ATPases that are only found in eukaryotes.

The P₁-type ATPases exist predominately in prokaryotes but are omnipresent across all domains of life⁴⁴: P_1A-ATPases are involved in K⁺ transport whereas P_1B-ATPases are important for maintaining transition metal homeostasis. P_1B-ATPases are known to transport Cu^{+ 45, 46}, Ag^{+ 47}, Zn^{2+ 48}, Cd^{2+ 49}, Cu^{2+ 50}, Co^{2+ 51} and Fe^{2+ 27, 36, 37}. The structure of a typical P_1B-ATPase includes a transmembrane domain with 6–8 helices, a soluble actuator domain, and an ATP-binding domain⁵² (Fig. 2). The P_1B-ATPases can be further divided into seven subclasses based on sequence similarity and metal substrate specificity⁵². The P_1B4-type ATPases were originally assigned a role in Co²⁺ export, based on the properties of some of the first characterized members⁵³. However, P_1B4-type ATPases have recently been found to function instead, or in addition, as Fe²⁺ efflux transporters including Bacillus subtilis PfeT³⁶, Listeria monocytogenes FrvA²⁷, Mycobacterium tuberculosis CtpD³⁷, and group A Streptococcus PmtA^{38, 39}.

PfeT in Bacillus subtilis

B. subtilis is a Gram-positive soil microorganism and encodes two transcriptional regulators critical for iron homeostasis, Fur_Bs and PerR. Fur_Bs is a global transcriptional regulator of iron homeostasis analogous to Fur_EC⁵⁴ and PerR mediates the adaptive response to peroxide stress by regulating genes involved in iron storage and peroxide detoxification²⁹. The regulons for both Fur_Bs and PerR have been well defined^{55, 56}. Fur_Bs senses intracellular iron sufficiency and represses genes that are involved in siderophore synthesis and uptake^{54, 57}. Fur_Bs also regulates an iron sparing response mediated by the small non-coding RNA FsrA (Fig. 1) and its coregulators FbpA, FbpB and FbpC^58–60. This system, analogous to RyhB in E. coli, blocks the translation of non-essential iron-containing enzymes such as aconitase and succinate dehydrogenase^58–60. PerR regulates peroxide detoxification enzymes (catalase, alkyl hydroperoxide reductase), iron sequestration (MrgA) and the P_1B4-type ATPase (PfeT). Although the Fur and PerR regulons are largely non-overlapping, pfeT is the exception and is regulated by both proteins²⁶. The result is that pfeT is induced by either peroxide stress or by iron excess (unpublished data, Pinochet-Barros A & Helmann JD).

PfeT is one of three P_1B ATPases encoded by B. subtilis. CopA is a P_1B1-ATPase that functions as a Cu⁺ efflux transporter and, appropriate to its function, is regulated by the CsoR Cu⁺ sensor⁶¹. CadA is a P_1B2-ATPase that confers resistance to Cd²⁺, Zn²⁺, and Co²⁺ and is regulated by the divalent cation sensor CzrA⁶². PfeT (formerly named as ZosA) is a P_1B4-type ATPase and was discovered as a transporter induced by H₂O₂ that plays a role in protecting cells against oxidative stress³⁵. Initial results indicated that deletion of pfeT enhanced Zn²⁺ tolerance, as monitored in cells lacking the CadA efflux system³⁵. This led to the proposal that PfeT might function as a Zn²⁺ importer under oxidative stress conditions, consistent with the idea that Zn²⁺ has a role in protecting cells against oxidative damage³⁵. As a result, PfeT was originally named for this proposed role as ZosA (Zn²⁺ uptake under oxidative stress)³⁵.

Contrary to this model, most P_1B-type ATPases function in metal export rather than import, which motivated a reinvestigation of the role of PfeT. Further study revealed that a pfeT null mutant is sensitive to Fe²⁺ and Fe³⁺, particularly under acidic media conditions, but not to Zn²⁺ or Co²⁺. Moreover, a pfeT null mutant accumulates elevated levels of intracellular Fe²⁺, as judged by sensitivity to the Fe²⁺-activated antibiotic streptonigrin and by direct chemical measurement³⁶. Biochemical studies confirmed that the ATPase activity of PfeT is induced the most by Fe²⁺, with modest induction by Co²⁺ but not with other metals, including Zn²⁺. In addition to H₂O₂, pfeT is strongly and specifically induced by iron, but not by other metals. Together, these findings indicate that PfeT function as a peroxide- and iron induced ferrous efflux transporter³⁶. The ability of PfeT to protect against H₂O₂ is secondary to that of the detoxification enzymes catalase and alkyl hydroperoxide reductase. However, PfeT plays a dominant role in protecting cells from iron overload with the MrgA miniferritin playing a secondary role³⁶. The revelation that PfeT functions in Fe²⁺ efflux, in turn, prompted a re-evaluation of the roles of several other P_1B4-type ATPases in bacterial iron homeostasis.

FrvA in Listeria monocytogenes

L. monocytogenes is the causative agent of the foodborne disease listeriosis, which is associated with central nervous system infections and bacteraemia. FrvA (Lmo0641) is a P_1B4-ATPase originally described as a Fur-regulated virulence factor⁶³. FrvA was proposed to function as a heme exporter that was suggested to be induced by iron deficiency and to be under negative regulation of both Fur and PerR^{63, 64}. However, a different transcriptome study showed a downregulation of frvA in a fur null mutant⁶⁵, indicating a positive regulatory role of Fur in frvA expression.

To resolve these contradictory reports of iron regulation, and to test if FrvA might function in Fe²⁺ efflux, the mutant phenotype was reinvestigated and the FrvA protein was purified for biochemical studies²⁷. As predicted based on studies of pfeT, a frvA null mutant was sensitive to iron intoxication, but not to other metals or heme. Like B. subtilis pfeT, frvA is positively regulated by Fur in response to high Fe²⁺ levels and is repressed by PerR^{27, 64}. Biochemical studies indicate the FrvA ATPase activity is stimulated most strongly by Fe²⁺ with weaker stimulation in the presence of Co²⁺ or Zn²⁺. Based on the Fe²⁺ concentration dependence of ATPase activity, FrvA seems to have a higher affinity for Fe²⁺ than B. subtilis PfeT. Consistent with this, not only does FrvA complement the iron-sensitive phenotype of a B. subtilis pfeT null mutant, its expression depletes the cytosol of iron (even under iron-rich conditions) thereby leading to derepression of the Fur regulon²⁷. These results support the hypothesis that FrvA functions as a Fe²⁺ efflux transporter that protects cells from Fe²⁺ intoxication²⁷.

FrvA is required for virulence in murine and insect (Galleria mellonella) infection models⁶³. The frvA null mutant strain shows strong attenuation in virulence, but is still able to invade and propagate inside antigen-presenting cells⁶⁶, suggesting an important link between iron homeostasis and virulence, but it is not clear at which stage(s) of the L. monocytogenes life cycle FrvA is important. The phagocytic vacuole is generally considered to be an iron-limited environment. One possibility is that the expression of high affinity iron uptake systems by iron limitation during infection or in the phagocytic vacuole can contribute to iron overload upon escape of cells into the relatively iron-rich cytosol. Alternatively, the imposition of oxidative damage from host immune cells may trigger iron release from listerial iron enzymes and this may lead to iron overload. The points in the infection cycle where FrvA plays a critical role are not yet clearly defined and provide an interesting avenue for future research.

CtpD in Mycobacterium tuberculosis

M. tuberculosis is an obligate pathogen and the causative agent of human tuberculosis. Nearly one-third of the world's population is infected with M. tuberculosis, which can persist in a latent state for decades and then later emerge (in ~10% of cases) as an active lung infection. M. tuberculosis encodes a total of 11 P-type ATPases, which have been suggested to be possible targets for therapeutic intervention⁶⁷. Of these, two encode P_1B4-ATPases: CtpD (Rv1469) and CtpJ (Rv3743)³⁷. CtpD, but not CtpJ, was found to be important for survival in macrophages and the mouse lung³⁷. Biochemical studies had previously highlighted the activity of these two P_1B4-ATPases with Co²⁺, but it was not clear why M. tuberculosis would encode two such proteins, nor was it understood why Co²⁺ efflux would be important for survival in the host.

In light of the finding that PfeT functions as an Fe²⁺ efflux transporter, the roles of CtpD and CtpJ were reinvestigated. Biochemical studies indicated that the ATPase activity of CtpD is most strongly activated by Fe²⁺. Although Co²⁺ also activates ATPase activity, the maximal activity (V_max) is 10-fold lower than with ferrous iron³⁷. CtpD also binds Fe²⁺ with 3-fold higher affinity than Co²⁺. In contrast, the CtpJ ATPase activity is activated by both Fe²⁺ and Co²⁺, and has a slightly higher affinity for Co²⁺ than Fe²⁺. To better understand their roles in vivo, metal accumulation and sensitivity was monitored for strains lacking either ctpD or ctpJ. The ctpD mutant strain did not accumulate Co²⁺ and was impaired in growth in iron-amended medium, consistent with a primary role in resistance to iron intoxication³⁷. Mutation of ctpJ led to a significant increase in Co²⁺ accumulation and expression was induced by Co²⁺, consistent with a primary role in Co²⁺ resistance³⁷. However, the ctpJ mutant was also growth impaired in the presence of excess iron. Thus, these two paralogous transporters seem to have overlapping metal selectivity, but largely distinct physiological roles. Further studies are needed to understand the molecular mechanism of substrate specificity, but based on X-ray absorption spectroscopy (XAS) analysis, it is likely that distinct metal coordination geometry plays an important role³⁷.

During infection, M. tuberculosis propagates in the host macrophages, which are considered iron-poor environments. Just as noted for L. monocytogenes, it is not yet clear where in the infection process cells experience iron intoxication. Further studies are needed to better understand the conditions that lead to induction of ctpD. In prior work ctpD was not induced by metals such as Co²⁺, Zn²⁺, and Ni²⁺, but its cognate substrate Fe²⁺ was not tested⁶⁸. It might be induced by Fe²⁺ and, by analogy with its orthologs, this might involve an iron-sensing transcription factor. IdeR, a member of DtxR family, is the major iron-dependent transcriptional regulator in M. tuberculosis^{10, 11}. IdeR represses transcription of genes involved in iron uptake and siderophore biosynthesis and activates expression of genes encoding iron-storage proteins such as bacterioferritin and a ferritin-like protein^{10, 11}. Since M. tuberculosis is primarily a pathogen of the mammalian respiratory system it might frequently encounter oxidative stress. Thus, it is also possible that ctpD might be induced in response to H₂O₂ stress. Future work to monitor the expression of ctpD in vitro in response to specific stresses and in vivo during the course of infection will be needed to elucidate the physiological role of CtpD during the infection process.

PmtA in group A Streptococcus

Group A Streptococcus (GAS), a human pathogen, is the causative agent of a wide range of diseases, from mild skin infection to life-threatening diseases such as necrotizing fasciitis⁶⁹. GAS encodes a P_1B4-type ATPase under regulation of PerR, and was therefore named a PerR-regulated metal transporter (PmtA). In a perR null mutant, high level expression of pmtA is associated with derepression of genes normally responsive to cellular zinc status due to repression by AdcR⁷⁰, a Zn²⁺-dependent repressor. This simplest interpretation of this result is that PmtA may function as a Zn²⁺ efflux transporter. Consistent with this notion, a perR null mutant has an increased resistance to Zn²⁺, and this depends on PmtA⁷⁰. However, it is unclear why cells would efflux Zn²⁺ in response to H₂O₂ stress, nor is there any evidence that PmtA is important for Zn²⁺ resistance in wild-type cells, which presumably relies on the Zn²⁺-inducible CzcD efflux pump to ameliorate Zn²⁺-toxicity.

By analogy with PfeT and its orthologs, an alternative interpretation is that the primary role of PmtA is as a H₂O₂-inducible Fe²⁺-efflux pump and that activity with Zn²⁺ may only be revealed when it is constitutively overexpressed in a perR null mutant. Two recent studies have confirmed the primary role of PmtA as an Fe²⁺-efflux pump^{38, 39}. PmtA is important for resistance to iron intoxication, and a pmtA null mutant accumulates elevated levels of intracellular iron. As expected, expression of pmtA is strongly induced by Fe²⁺. Although a pmtA null mutant shows similar sensitivity to peroxide stress as a wild type strain in the absence of excess Fe²⁺, it exhibits significantly increased susceptibility to peroxide stress when treated with Fe²⁺. Since GAS is catalase negative, PmtA might be a frontline defense against peroxide stress. PmtA is also a critical virulence factor and is required for survival during infection in both intramuscular and subcutaneous mouse models³⁸, which again links iron efflux and peroxide resistance to pathogen virulence.

Nia in Sinorhizobium meliloti

In addition to the P_1B4-ATPases featured above, it is possible that P_1B-ATPases of other groups may also have physiologically relevant activity with iron. One example is Nia, a P_1B5-ATPase with a C-terminal hemerythrin domain. Since hemerythrin domains bind O₂ via a diiron active site, this suggests a possible role in O₂-sensing^{71, 72}. Nia is encoded by the symbiotic plasmid A of Sinorhizobium meliloti, a nitrogen fixing microbe in the Rhizobiales lineage that has a symbiotic relationship with legumes in which it establishes nodules associated with roots.

Consistent with a possible role in Fe²⁺ efflux, a nia null mutant accumulates Fe²⁺ under excess metal conditions⁷³. However, Nia also functions with Ni²⁺ and a nia null mutant accumulates Ni²⁺ when in excess. The precise physiological role of Nia is not yet resolved. Biochemical assays suggest that Nia interacts with both Fe²⁺ and Ni²⁺ (but not Co²⁺). However, a nia null mutant showed moderate sensitivity to Ni²⁺, but not to Fe²⁺, under the conditions tested⁷³. Expression of nia was moderately induced by Fe²⁺ (3-fold), Ni²⁺ (3-fold), and Co²⁺ (2-fold), but not by other metals. Interestingly, nia was most strongly induced (20-fold) in root nodules, thought to be a microaerobic, iron-rich environment⁷⁴. These results lead to a model in which Nia is expressed in nitrogen-fixing root nodules, in response to either iron excess or microaerobic conditions. The C-terminal hemerythrin domain may also participate in or regulate transport activity, perhaps in response to O₂⁷³. More work needs to be done to characterize the details of nia gene regulation and to more clearly define the physiological role of Nia during the S. meliloti -plant symbiosis.

Cation diffusion facilitator (CDF) proteins

Cation diffusion facilitators (CDFs) are a family of membrane-bound proteins that export and thereby confer tolerance to heavy metal ions^{75, 76}. CDF proteins are ubiquitous in bacteria, archaea, and eukaryotes⁷⁷. Collectively, bacterial CDF proteins have been implicated in transport of a wide range of metal ions (Zn²⁺, Cd²⁺, Co²⁺, Ni²⁺, Fe²⁺ and Mn²⁺) with some transporters able to transport multiple metals^78–82. Phylogenetic analysis of the CDF transporters defines three major groups corresponding to substrate specificity: 1) manganese efflux, 2) iron/zinc efflux, 3) zinc and other metals (but not manganese or iron) efflux⁸³.

A typical bacterial CDF contains an N-terminal domain (NTD), 6 transmembrane helices (TM), a histidine-rich interconnecting loop (IL) between TM4 and TM5, and a C-terminal cytoplasmic domain (CTD)⁷⁵ (Fig 2). However, the detailed mechanisms of metal selectivity are unknown. Some studies suggest the cytoplasmic domain or the IL loop is important for metal specificity^84–86, but other studies highlight the role of residues in the TM3 helix on metal selectivity⁸⁷. For the E. coli FieF transporter, evidence supports a role for a tetrahedral metal-binding site formed between TM2 and TM5 in metal selectivity⁸⁸. So far, there is no unifying model that can account for metal selectivity of CDF proteins.

FieF in E. coli: Zn²⁺ vs. Fe²⁺ efflux

There are two CDF transporters in E. coli: ZitB and FieF (also named as YiiP). ZitB is the secondary zinc efflux system that is critical for maintaining zinc homeostasis only when the zinc efflux ATPase ZntA is absent⁸⁹. FieF has been studied for more than a decade, but its physiological function has been controversial. In 2004, the first two reports of its structural analysis were built on the assumption that FieF acts as a zinc efflux protein^{90, 91}. In fact, prior studies had demonstrated that fieF is induced by either zinc or iron⁸⁹. However, ectopic expression of FieF does not restore zinc tolerance in a zinc-sensitive strain, suggesting it might not play a role in zinc homeostasis⁸⁹.

Physiological studies suggest that the major physiological role of FieF may be in iron tolerance. Indeed, FieF is important for full resistance to iron intoxication in a fur null mutant, where iron homeostasis is disrupted and iron uptake systems are constitutively expressed⁹². Ectopic expression of FieF leads to reduced accumulation of iron in a fieF null mutant. Moreover, reconstitution of FieF in proteoliposomes showed that it mediates iron transport in vitro⁹². These results all support the assignment of FieF (ferrous iron efflux) as an iron efflux transporter. However, this notion has been challenged by others. For example, FieF was shown to selectively bind zinc and cadmium with high affinity, but not iron or other metals tested⁹³. Based on the site-directed fluorescence resonance energy transfer (FRET) measurements, Lu et al. proposed an autoregulation model of transport activity in response to intracellular zinc levels⁹⁴. Currently, FieF (YiiP) is referred to as a Zn²⁺ transporter in most published papers.

Ever since its structure was solved in 2007⁷⁹, FieF has been considered as a prototype for bacterial CDF proteins, which makes it more frustrating that its physiological role has remained controversial. The regulation of fieF expression has not been well defined, but it does not appear to be regulated by Fur⁹². The physiological studies of FieF are certainly supportive of a role in Fe(II) efflux. This inference is further supported by the observation that the FieF homologs MamM and MamB form a heterodimeric CDF protein required for Fe(II) import into vesicles in support of magnetosome formation in the magnetotatic bacterium Magnetospirillum gryphiswaldense^{95, 96}.

AitP in Pseudomonas aeruginosa

Pseudomonas aeruginosa is Gram-negative, opportunistic pathogen that is highly antibiotic resistant. P. aeruginosa encodes three paralogous CDF efflux systems: CzcD (PA0397), AitP (PA1297), and YiiP (PA3963). Of these, the alternative iron transport protein (AitP) most likely functions physiologically in Fe²⁺ efflux. Deletion of aitP leads to an increased sensitivity to both Fe²⁺ and Co²⁺, increased intracellular accumulation of both ions, and decreased survival in presence of H₂O₂⁹⁷. The observed sensitivity to H₂O₂ is most consistent with a role in Fe²⁺ efflux, as noted above for P-type ATPases. In contrast with AitP, the CzcD and YiiP proteins were inferred to function physiologically in Zn²⁺ resistance, although this role is largely masked in wild-type cells by the activity of the Zn²⁺ efflux P-type ATPase, ZntA⁹⁸. All the three transporters are critical for virulence in a plant infection model⁹⁷. However, it remains unclear why this organism requires multiple classes of Zn²⁺ efflux proteins or under what conditions the three proteins are physiologically important during the infection process.

FeoE in Shewanella oneidensis MR-1

Shewanella oneidensis MR-1 is a facultative anaerobe in the γ-proteobacterium family that is capable of respiration using metals (e.g. manganese, lead, uranium and ferric iron) as electron acceptors⁹⁹. S. oneidensis cells are usually pink or red, reflective of a high iron content in hemoproteins and cytochromes¹⁰⁰. When Fe³⁺ is used as a terminal electron acceptor, cells generate a large amount of soluble Fe²⁺ which could potentially lead to iron intoxication. FeoE, a CDF protein, is required for cell growth during anaerobic iron respiration, and deletion of feoE increased susceptibility to Fe²⁺ intoxication, consistent with a physiological role in Fe²⁺ efflux¹⁰¹. Further work is required to understand how feoE expression is regulated. It is unclear, for example, whether feoE is induced in response to excess iron. Fur is the primary regulator that modulates iron acquisition in S. oneidenis¹⁰², and is a candidate for an iron-responsive transcription factor that could be involved.

Major facilitator superfamily (MFS)

The major facilitator superfamily (MFS) of membrane transporters function with a wide scope of small molecules such as ions, nucleosides, amino acids, small peptides, and lipids¹⁰³. They can be categorized into three groups: uniporters that transport a single substrate, symporters that transport a substrate coupled with another ion (generally a proton), and antiporters that transport two substrates in opposite directions^{104, 105}. All the MFS transporters share a canonical structural fold composed of two distinct domains [Fig. 2], each consisting of six transmembrane helices. The substrate binding site is located at the interface between these two domains¹⁰³.

The mechanism of transport by MFS proteins is not clear, but several related models have been proposed. The first, an alternate-access model, was proposed more than five decades ago¹⁰⁶. This model speculates that the transporters undergo a conformational change that alternates between a form where substrate can bind from one side of the membrane to one where it can only bind from the other side. This has been validated by many structural studies such as the xylose/H⁺ symporter XylE and for LacY^107–109. The second, a rocker-switch model, postulates that conformational changes are accomplished through rocker-switch-type rotation between the N and C domain. This model is supported by some open-conformation structures¹¹⁰ but not by the structures in occluded states^111–114. A third, clamp-and-switch model, provides a two-step transport mechanism: a clamping step that mediates occlusion of the binding site and a switching step that mediates the exposure of the binding site. This model postulates four conformational states: inward open, outward open, inward-facing occlusion, and outward-facing occlusion¹⁰⁵. This model is in a good agreement with studies of some MFS transporters^{115, 116}, but more structural analyses combined with biochemical and computational analyses are needed to further understand the transport mechanism of MFS transporters.

IceT (iron and citrate efflux transporter) in Salmonella Typhimurium

Salmonella Typhimurium is a Gram-negative pathogen commonly found in the gastrointestinal tract. IceT (MdtD) is a member of the MFS superfamily in S. Typhimurium. The mdtABCD baeSR operon encodes IceT and two other systems: a RND (resistance-nodulation-division) drug efflux system MdtABC and a two-component regulatory system BaeSR that regulates antibiotic resistance and efflux^117–119. IceT is proposed to be an iron-citrate efflux transporter and it can export either iron citrate or citrate alone¹²⁰. The iceT null mutant shows increased susceptibility to the antibiotic streptonigrin (SN), the activity of which is modulated by the level of intracellular free iron¹²¹. This result suggests that the mutation of iceT leads to an increase in intracellular labile iron pools. Consistent with this result, induction of IceT expression leads to reduced levels of intracellular iron¹²⁰.

Although the mdtABCD baeSR operon is not induced directly by high Fe^{2+ 122}, it is induced by disruption of iron homeostasis in a fur null mutant where iron uptake systems are constitutively expressed, supportive of a physiological role for IceT in iron efflux. Although IceT confers resistance to peroxide stress in a fur null mutant, the mdtABCD baeSR operon is not induced by H₂O₂ or superoxide-generating reagents such as paraquat¹²⁰. However, it is induced by nitric oxide, which is also known to interact with the labile iron pool¹²⁰. The significance of the regulation of IceT, together with its co-transcribed ABC transporter, by the BaeSR two-component system is not understood, nor is it yet clear whether or not IceT is important for pathogenesis.

Membrane bound ferritin A (MbfA) in Agrobacterium tumefaciens and Bradyrhizobium japonicum

Agrobacterium tumefaciens belongs to the Rhizobiales lineage and is the causative agent of the economically important plant disease, crown gall. MbfA was originally described as membrane-bound ferritin A, and is a member of the erythrin-vacuolar iron transport (Er-VIT1) ferritin-like superfamily. MbfA has two major domains: an N-terminal ferritin-like or Er domain (Er) and a C-terminal membrane-embedded vacuolar iron transporter domain (VIT1) (Fig. 2). The Er domain has a di-iron binding site and the VIT1 domain shows sequence homology to Arabidopsis VIT1, which is responsible for transferring iron into vacuoles¹²³. Ferritin is a cytosolic iron storage protein ubiquitous in prokaryotes and eukaryotes¹²⁴, however, MbfA is not a bona fide ferritin and its physiological function was not immediately apparent.

Plant hosts often produce reactive oxygen species as a defense mechanism in response to microbial infection. Initial studies revealed that MbfA confers resistance to H₂O₂ stress, suggesting that it may play an important role in plant-pathogen interaction¹²⁵. Moreover, mbfA expression was induced in response to high iron conditions as sensed by the iron response regulator protein, Irr¹²⁵. However, these results could not distinguish between a role for MbfA in sequestration of iron (through its ferritin domain) or iron efflux. A follow up study revealed that MbfA is important for resistance to iron intoxication under acidic conditions (pH 5.5), which enhances iron solubility thereby promoting toxicity¹²⁶. Compared to wild-type, an mbfA null mutant had a modest increase in intracellular total iron as well as labile iron¹²⁵. Since its expression is induced by high iron under acidic conditions¹²⁵, and leads to reduced intracellular iron levels, MbfA was postulated to function as an iron efflux transporter¹²⁵.

Bradyrhizobium japonicum also encodes an MbfA protein implicated in iron efflux¹²⁷. B. japonicum is a nitrogen-fixing endosymbiotic microbe that, like A. tumefaciens, belongs to the Rhizobiales lineage. As in A. tumefaciens, iron homeostasis in B. japonicum is also under control of Irr¹²⁸, which regulates iron uptake, storage, and utilization¹²⁹. MbfA in B. japonicum is specifically induced by high iron and confers resistance to iron intoxication and H₂O₂ stress. Moreover, an mbfA null mutant accumulates significantly high levels of iron.

Collectively, these data support the idea that MbfA functions physiologically as an iron efflux transporter¹²⁷. Interestingly, the N-terminal ferritin-like domain located on the cytoplasmic side of inner membrane is required for iron transport activity and stress resistance. The purified ferritin domain forms a dimer in solution, which suggests that MbfA may dimerize to form a functional channel¹²⁷. By mediating the efflux of Fe²⁺, MbfA functions cooperatively with bacterioferritin (Bfr), which functions in iron sequestration, to prevent iron intoxication¹³⁰. Mutation of either mbfA or bfr increases Fe²⁺ sensitivity, but a double mbfA bfr mutant is extremely sensitive to iron¹³⁰.

Conclusions

Efflux systems play a central role in the resistance of bacteria to heavy metals, but their role in iron homeostasis has been relatively slow to emerge. This is perhaps a reflection of the fact that iron limitation is a far more prevalent challenge for bacteria than iron intoxication¹³¹, due in part to the very low solubility of iron under aerobic conditions of near neutral pH. Recent results, however, have greatly expanded our appreciation of the central importance of iron efflux systems and their contribution to virulence in human pathogens^{27, 37, 38}. This implies that iron intoxication imposes a selective pressure during infection, although how this arises is not yet clear. For example, iron intoxication may arise from an uncontrolled influx of iron into the cell from the outside. Indeed, it is thought that macrophages impose Zn²⁺ and Cu⁺ toxicity on engulfed bacteria by import of metals into the phagolysosome¹³². However, iron is not known to be imported into the phagocytic vacuole. Iron overload may also result when bacteria exposed to an iron limited environment, and therefore expressing high affinity uptake systems, transition to an iron-rich environment. The sudden influx of iron may then be best accommodated by storage or efflux. Alternatively, or in addition, iron intoxication may arise from within the cell. For example, oxidative stress may lead to the release of iron from abundant iron-sulfur and mononuclear iron enzymes, thereby leading to an increase in cytosolic iron levels.

Iron intoxication may also be present in specific environments. For example, acidophilic bacteria grow in low pH environments where iron concentrations may be 10¹⁸ times higher than that found in pH neutral environments¹³³. In the case of iron-respiring bacteria, high local concentrations of Fe²⁺ may be produced by reduction of Fe³⁺-containing minerals¹⁰¹. Further work is needed to better define the prevalence of iron intoxication in natural environment settings and the role of iron efflux in these environments.

With the identification of the several families of iron efflux systems noted here, the stage is now set for further structural, biochemical and genetic studies to address their mechanisms of metal selectivity. It is presently unclear how these efflux transporters discriminate Fe²⁺ from competing substrates and how, at a structural level, efflux is coupled to substrate binding and energy consumption. It is also unclear why some cells rely on ATP-dependent P-type transporters and others utilize CDF proteins, which are coupled to the proton motive force. It is notable that in several cases efflux pumps were initially assigned a role for substrates others than Fe²⁺ (PfeT, FrvA, CtpD), and in other cases (FieF, Nia) the most relevant physiological substrate is still unclear. This highlights the fact that metal selectivity cannot be easily predicted from protein sequence alone, and biochemical assays need to be interpreted in context of the physiology of the organisms. In several of the cases described, the most compelling evidence to assign function has emerged from a careful analysis of mutant phenotypes combined with detailed analysis of regulation to infer those conditions that specifically induce expression.