Prokaryotic Ferrous Iron Transport: Exploiting Pools of Reduced Iron Across Multiple Microbial Environments

Abstract

Iron is essential to nearly all forms of life, but the redox activity of this element necessitates its cellular regulation. All iron-utilizing organisms require this nutrient to be tightly managed, which is accomplished by a suite of proteins and nucleic acids involved in the acquisition, the delivery, the storage, the regulation, and the export of this essential element. For infectious organisms, iron acquisition systems are commonly associated with intracellular growth, survival, and virulence of pathogens, which are dynamically able to modify their iron uptake strategies in response to changing host environments. The past few decades have been marked with an increased understanding of pathogenic ferric iron (Fe³⁺) and heme acquisition. In contrast, though the necessity of ferrous iron (Fe²⁺) transport for pathogenesis has been established, the precise details of this process remain enigmatic, and Fe²⁺ transporters remain unexploited as drug targets to combat drug-resistant organisms. This chapter will overview a current understanding of Fe²⁺ transport in microbes and highlight gaps in our knowledge that must be closed in order to establish a comprehensive understanding of unicellular Fe²⁺ metabolism.

1 Introduction

Throughout evolution, iron has emerged to become one of the most vital elements to virtually every living organism. In the modern era, biological systems that accomplish a suite of cellular processes, from aerobic cellular respiration, N₂ fixation, gene regulation, to even DNA biosynthesis make use of iron in all of its different forms (Sestok et al. 2018; Andrews et al. 2003; Lau et al. 2016). Iron has a diverse functionality and may be utilized as an ionic cofactor and bound by biological macromolecules, complexed within iron-sulfur clusters, or even chelated by organic molecules such as protoporphyrin-IX (heme) (Andrews et al. 2003; Sestok et al. 2018). Biology is replete with essential proteins and enzymes that use some or even all forms of this element. For example, iron-dependent ribonucleotide reductases reduce ribonucleotides to deoxyribonucleotides by making use of a diiron center that shuttles electrons for nucleotide reduction (Torrents 2014). Nitrogenase enzymes, which allow bacteria to fix N₂, utilize multiple iron-sulfur clusters ([4Fe-4S], [8Fe-7S], and [7Fe-Mo-9S]) throughout the fixation cycle (Hu and Ribbe 2015). Hemoglobin (Wittenberg et al. 2002) and myoglobin (Brunori 2010) both require a heme b cofactor for O₂ transport. This large assortment of iron-containing proteins clearly demonstrates the dependence on iron for the survival of organisms found across every kingdom life.

Iron is capable of fulfilling such a dynamic role in life because of its abundance on earth and its versatility as an element, if it can be acquired. For example, iron is the most abundant transition metal on earth’s crust, comprising up to ca. 5% of the outermost layer of the planet. However, unlike oxygen, silicon, and aluminum, which make up significantly more of the outer layer of Earth, the chemistry of iron is much more diverse. Iron can sample multiple oxidation and spin states, and its electronic reactivity can be modulated by the ligands responsible for binding the metal in biological macromolecules. This tunability controls the redox potential of iron, which can span from ca. −300 mV to +700 mV (i.e., a 1 V range), affording organisms the potent oxidizing and reducing iron-based cofactors necessary to accomplish a wide swath of complex chemical transformations (Andrews et al. 2003). However, despite its abundance, acquisition of this essential element is no small feat. While Fe³⁺ (ferric iron) predominates in oxic environments, it is highly insoluble (ca. 10⁻¹⁸ M at pH 7.0) and easily forms recalcitrant ferric oxides and hydroxides, whereas Fe²⁺ (ferrous iron) is more soluble (as high as 0.1 M at pH 7.0) but is present under anoxic and reducing conditions and is susceptible to rapid oxidation (Fig. 12.1) (Andrews et al. 2003; Krewulak and Vogel 2008; Winterbourn 1995), presenting a conundrum to many organisms. Given the critical importance of iron to life, decades of research have focused on the mechanisms of iron transport, utilization, and even regulation. Much of this research has centered around bacterial iron homeostasis due to its simpler mode of study and spurred by the strong relationship between iron availability and bacterial pathogenesis.

Fig. 12.1

The Fenton reaction. In the presence of hydrogen peroxide, ferrous iron (Fe²⁺) is readily oxidized into ferric iron (Fe³⁺), forming both a hydroxyl radical and a hydroxide ion in the process. Hydroxyl radicals can lead to harmful and deleterious oxidation of biomolecules if uncontrolled

In order to utilize iron, this nutrient must first be obtained from the environment and subsequently transported to the necessary locations within the cell where the metal is needed. This process requires organisms to employ diverse iron sequestration systems capable of functioning under a variety of different physiological niches and in different environments to exploit whatever pools of iron may be available (Andrews et al. 2003; Sestok et al. 2018). In Gram-negative bacteria, Fe³⁺ can be scavenged by siderophores, low molecular weight molecules with a high affinity for Fe³⁺ (K_aff > 10³⁰ M⁻¹) that are secreted into the extracellular space. Siderophores compete for Fe³⁺ already bound to host proteins and help to keep Fe³⁺ soluble during the transport process. Uptake of the Fe³⁺-siderophore complex from the extracellular space into the periplasm typically occurs via a TonB-dependent outer membrane receptor coupled to the proton motive force (PMF) and often specific to the siderophore (Andrews et al. 2003; Sestok et al. 2018; Krewulak and Vogel 2008; Chu et al. 2010; Ellermann and Arthur 2017). Periplasmic binding proteins then chaperone the complex to a dedicated inner membrane protein, typically an ATP-binding cassette (ABC) transporter. Once in the cytoplasm, Fe³⁺ can be reduced to Fe²⁺ or the siderophore can be degraded to release the iron (Cain and Smith 2021). Uptake of siderophores is similar in Gram-positive bacteria, except that the Fe³⁺-siderophore complex binds to a membrane anchored lipoprotein before being handed off to the ABC transporter (Andrews et al. 2003; Sestok et al. 2018; Krewulak and Vogel 2008; Chu et al. 2010; Ellermann and Arthur 2017). Heme sequestered from host proteins also serves as a form of iron to bacteria; however, because of its dramatically different composition from that of free iron, dedicated heme acquisition and transport systems are necessary. Pathogenic bacteria can employ hemolysins and proteases to release heme from red blood cells. Bacteria can then bind heme by using dedicated hemophores or by utilizing outer membrane receptors to transport host heme-binding proteins into the cell (Andrews et al. 2003; Sestok et al. 2018; Krewulak and Vogel 2008; Huang and Wilks 2017; Richard et al. 2019). The heme-hemophore complex is transported across the outer membrane in a TonB-dependent manner, similar to siderophore uptake. ABC transporters may also shuttle hemophores across the cytoplasmic membrane where the heme is degraded by heme oxygenase (HO), releasing iron and protoporphyrin catabolites (Andrews et al. 2003; Sestok et al. 2018; Krewulak and Vogel 2008; Huang and Wilks 2017; Richard et al. 2019). Both ferric iron and heme serve as indispensable iron sources for bacteria and have been well studied.

Recently, it has become apparent that Fe²⁺ acquisition and utilization are both necessary for bacterial growth, survival, and in some cases, for pathogenesis (Sestok et al. 2018; Andrews et al. 2003). However, bacterial ferrous iron uptake is not as well characterized as that of ferric iron and heme uptake. Ferrous iron is likely to be the dominant iron source for bacteria that thrive in microaerophilic or anoxic environments, such as those living within biofilms, and those living in acidified, reducing environments like the stomach. Additionally, there is evidence that Fe³⁺ utilization is linked to Fe²⁺ transport, especially in cases where organisms utilize non-native siderophores known as xenosiderophores (Cain and Smith 2021; Schröder et al. 2003). These molecules can be transported across the outer membrane through TonB-dependent receptors specific for the xenosiderophores or through outer membrane receptors with broad specificity for several siderophores (Cornelis 2010; Sheldon and Heinrichs 2015). However, since organisms that can utilize xenosiderophores often lack siderophore-specific, inner membrane transporters, they must therefore either reduce Fe³⁺ to Fe²⁺, through the use of a ferric iron reductase (Cain and Smith 2021), or degrade the xenosiderophore within the periplasm. In either scenario, transport of Fe²⁺ into the cytoplasm requires dedicated and selective Fe²⁺ transporters. Ferric iron reductases can also function extracellularly and exhibit broad substrate specificity for different Fe³⁺-chelates, necessitating Fe²⁺ transport across the outer membrane. The process of Fe³⁺ reduction may involve a flavin cofactor (soluble ferric reductases) or heme b (membrane ferric reductases) to transport electrons for iron reduction (Cain and Smith 2021). Thus ferric iron reductases allow for organisms to generate and to utilize Fe²⁺ even in oxic environments (Schröder et al. 2003). Ferrous iron transporters have been discovered in many kingdoms of life, but most remain under- or poorly-studied. This chapter focuses on how microbes specifically exploit the presence of pools of reduced, ferrous iron in the environment to fulfill metabolic needs and to establish infection within host niches. Importantly, while reduced iron availability has changed throughout Earth’s evolution, microbes have adapted and maintained the ability to acquire this necessary nutrient through conserved mechanisms across multiple microbial environments.

2 Iron and Bacteria on Early Earth

Bacteria first appeared on Earth approximately 4 billion years ago and thrived in an environment remarkably different from that of today’s (DeLong and Pace 2001). The presence of banded iron formations (BIFs) have shown that Earth’s first oceans were ferruginous (rich in iron), and these mineral “snapshots” offer insight into how the composition of Earth’s oceans have changed throughout Earth’s evolution (Li et al. 2018; Schad et al. 2019; Konhauser et al. 2017). While there are multiple mechanisms by which iron oxidation to promote BIFs could have transpired, one prevailing hypothesis suggests formation occurred through oxidation of iron by bacteria, such as Gallionella ferruginea, as fossilized bacteria have been discovered co-deposited within BIFs (Konhauser et al. 2017). These ancient bacteria were micro-aerophilic, being either (or both) nitrate-reducing, phototrophic ferrous iron (Fe²⁺)-oxidizers or ferric iron (Fe³⁺)-reducers (Schad et al. 2019). Studies have shown that bacteria inhabiting the surface of the early oceans likely utilized Fe²⁺ oxidation to produce Fe³⁺-containing minerals as protective mechanisms against harmful UV-radiation prior to the more modern development of the ozone layer (Gauger et al. 2015, 2016). Fe²⁺ was also proposed to be involved in the photosynthetic process known as photoferrotrophy, whereby UV-radiation facilitated oxidation of dissolved Fe²⁺ species that was subsequently coupled to several essential assimilatory processes, such as reduced carbon fixation, CO₂ fixation, and even N₂ fixation (Konhauser et al. 2017; Schad et al. 2019). However, changes in Earth’s atmospheric composition quickly altered metal speciation on the surface and within the oceans.

The presence of significant levels of atmospheric O₂ during the Great Oxygenation Event (GOE) was thought to have appeared around 2.4 billion years ago, though this estimation is complicated by the lack of consensus as to exactly when and where cyanobacterial photosynthesis began (either in the early Archean Eon or the late Paleoproterozoic Eon) (Schad et al. 2019; Glass 2015; Shih 2019). Regardless of the precise date, the emergence of photosynthetic cyanobacteria resulted in the release of substantive amounts of O₂ into the atmosphere and dissolved within the oceans. This increase in O₂ caused dramatic changes in early microbial environments, leading to pressures on anaerobic metabolizers and the alteration of metal bioavailability, chiefly due to O₂’s effects on oxidation state speciation of multiple essential transition metals.

Notably and especially affected was iron, as the GOE heavily altered the iron biogeochemical cycle (Glass 2015; Schad et al. 2019; Shih 2019). As the distribution of iron shifted from the soluble Fe²⁺ form to the insoluble Fe³⁺ form, microbes were pressured to adapt to fulfill their metabolic needs or perish. As fossil records do not indicate a mass bacterial extinction event, it is clear that microbes overcame this challenge through evolved mechanisms of metal scavenging, metal storage, metal substitution, and a metabolic shift towards different metal speciation (Glass 2015). However, not every planetary niche was transformed by the GOE. Studies suggest that some ferruginous environments were preserved, and microbes in those environments likely maintained a necessity for Fe²⁺, which was still present in relatively high concentrations in these locations (Derry 2015; Konhauser et al. 2017).

Despite its difficulty to obtain from the environment, iron still serves as an indispensable metal cofactor for many enzymes (Glass 2015); however, before iron may be incorporated into proteins and enzymes and used in cofactor biosynthesis, iron must first be solubilized and transported into the cell. As microbes may encounter multiple forms of environmental iron, it is unsurprising that this process is accomplished by a variety of iron acquisition systems. A large number of microbial iron uptake mechanisms have been discovered and characterized extensively. Consistent with the essential nature of iron as a micronutrient, these acquisition systems have been determined to be essential for microbial growth, survival, and even virulence. However, most studies have focused on mechanisms of Fe³⁺ acquisition, such as Fe³⁺-siderophore transport, heme transport, and even the acquisition of host Fe³⁺-binding proteins from the extracellular space. In stark contrast, much less is known about Fe²⁺ uptake despite the importance of Fe²⁺ for the survival of both the first organisms that inhabited the planet and numerous bacteria located in modern-day ferruginous environments.

3 Fe²⁺ Acquisition Systems

Several Fe²⁺ uptake systems have been identified in bacteria, and it is likely that more exist that have not yet been characterized (Fig. 12.2). Furthermore, while some transport systems are specific to Fe²⁺ transport (such as the Feo and Efe systems), other systems mediate the transport of multiple divalent metal ions such as Mn²⁺, Cu²⁺, Zn²⁺, Ni²⁺, and Co²⁺. Given the different coordination properties of Fe²⁺ and Fe³⁺, and the different environments in which these two ions are obtained, different machinery is required for the transport of Fe²⁺ and Fe³⁺. For instance, siderophores typically coordinate Fe³⁺ in an octahedral geometry with hexadentate ligands and are highly specific for hard Lewis acids (Andrews et al. 2003; Krewulak and Vogel 2008). Such a molecule does not coordinate the softer Lewis acid Fe²⁺ with the same affinity or even the same geometry in many cases. Additionally, because Fe²⁺ is much more soluble than Fe³⁺, it is unnecessary for the ion to be bound to large chelates during the transport process but may instead be transported as an ion via Fe²⁺-specific transporters. Though not an exhaustive list, some of the best characterized Fe²⁺ transporters are described below.

Fig. 12.2

A cartoon of the currently known bacterial ferrous iron uptake systems, depicted here for a Gram-negative bacterium. Fe²⁺ is transported into the periplasm via an unidentified outer membrane transporter where it is then taken up by one or multiple inner membrane Fe²⁺ transport systems. MntH, ZupT, and SitABCD transport multiple divalent metal ions in addition to H⁺, while EfeUOB is specific for Fe²⁺, but is only functional in pathogenic bacteria. IroT/MavN has been identified in Legionella species and functions at the LCV membrane. In contrast, Feo is the most widespread, dedicated Fe²⁺ transport system in bacteria. Figure created with BioRender

3.1 Natural Resistance-Associated Macrophage Proteins (NRAMPs) and Divalent Metal Transporters (DMTs)

Natural resistance-associated macrophage proteins (NRAMPs) are a family of divalent metal transporters (DMTs), also known as solute carriers (SLCs), involved in both eukaryotic and prokaryotic iron homeostasis. Eukaryotes express two separate NRAMPs, NRAMP1 (or SLC11A1) and NRAMP2 (or SLC11A2/DMT1). NRAMP1 is phagosomally-expressed and aids in the fight against pathogens by sequestering iron (among other transition metals) from the invading bacterium during phagocytosis by macrophages, whereas NRAMP2 facilitates the absorption of dietary iron and is critical in maintaining iron homeostasis (Ehrnstorfer et al. 2014; Bozzi et al. 2016). Bacterial NRAMP homologues constitute three “clades” (A, B, C) and display a high degree of sequence conservation to their eukaryotic counterparts, suggesting that their metal specificity and transport mechanisms may be similar to those exhibited in eukaryotes (Ehrnstorfer et al. 2014; Bozzi et al. 2016).

Recent studies have sought to understand the mechanism of NRAMP-mediated divalent metal ion transport through a combination of structural and biochemical techniques. In an X-ray crystal structure of a truncated Staphylococcus capitis DMT (ScaDMT) (belonging to clade C but lacking 41 N-terminal residues) the transmembrane (TM) I region contains an inverted repeat of the first five TM helices (known as a LeuT fold) and an unstructured region in the first helix in the inverted repeat (between helix 1 and helix 6), providing the coordinating ligands for metal substrates (Ehrnstorfer et al. 2014). The metal-coordinating residues of a metal-bound state were initially determined by solving the X-ray crystal structure of Mn²⁺-bound ScaDMT (PDB ID: 4WGW), which revealed the side chains of Met, Asp, and Asn (conserved residues in human DMT1) as well as the peptidyl carbonyl oxygen all bound to Mn²⁺ (Fig. 12.3) (Ehrnstorfer et al. 2014). Further structural and functional characterization of ScaDMT demonstrated binding of Fe²⁺, Co²⁺, Ni²⁺, Cd²⁺, Pb²⁺, and Cu²⁺ (albeit in a slightly shifted position). Mn²⁺, Co²⁺, Ni²⁺, and Cd²⁺ could be transported into ScaDMT-containing liposomes, although Fe²⁺ transport was not confirmed in this study (Ehrnstorfer et al. 2014). Intriguing follow-up studies demonstrated that the coordinating Met lends itself importantly to specificity by discriminating against alkali earth metals (Bozzi et al. 2016), which could be a common theme amongst Fe²⁺ transporters.

Fig. 12.3

X-ray crystal structure of ScaDMT bound to Mn²⁺ (PDB ID: 4WGW). Mn²⁺ (purple sphere) is coordinated by side chains from Met, Asp, and Asn residues, and a peptidyl oxygen from Ala (expanded region). ‘N’ and ‘C’ represent the N- and C-termini respectively. Figure created with BioRender

An X-ray crystal structure of Deinococcus radiodurans NRAMP (DraNRAMP), a member of prokaryotic clade A NRAMPS, similarly displayed a LeuT fold in an inward-facing conformation but included the missing N-terminal residues from the ScaDMT structure. The position of TM helix 1a to that of 1b is striking as the angle between the two helices is 103°. This bent conformation is believed to create an “aqueous vestibule” that may allow for a water molecule to coordinate metal bound between helices 1 and 6 (Bozzi et al. 2016). Furthermore, molecular snapshots obtained from cysteine accessibility scanning suggest that DraNRAMP undergoes a series of conformational changes to facilitate the transport of divalent metal ions through an extracellular metal-permeation pathway before reaching the metal-binding site, which could be a common mechanism among NRAMPs (Bozzi et al. 2016). While more structural information will be necessary to elucidate the full mechanism of divalent metal transport in these proteins and to characterize mutations resulting in disease states, the current structures and functional analyses have contributed to understanding how a broad range of metal substrates can be recognized, demonstrating the high flexibility and substrate promiscuity of some transporters. These models could prove useful for other uncharacterized transporters.

The H⁺-dependent manganese transport system (MntH) (Fig. 12.2, pink) is a NRAMP homolog first identified in E. coli in 1969 (Silver and Kralovic 1969; Silver et al. 1970; Bhattacharyya 1970; Makui et al. 2000). Though MntH has a preference for Mn²⁺, it has also been linked to the transport of Fe²⁺, and cells overexpressing MntH accumulate Fe²⁺ via active transport, as observed by in vivo ⁵⁵Fe²⁺ uptake experiments. In this study, ⁵⁵Fe²⁺ uptake could also be inhibited by Mn²⁺ and Zn²⁺ when supplied at a 100-fold molar excess or by the addition of protonophores (Makui et al. 2000). Similar experiments performed to monitor in vivo ⁵⁴Mn²⁺ uptake reveal the same proton-dependent uptake as ⁵⁵Fe²⁺. However, ⁵⁴Mn²⁺ accumulated to approximately ten-fold higher concentration over ⁵⁵Fe²⁺, demonstrating MntH’s specificity for Mn²⁺ (Makui et al. 2000). In addition, the overexpression of MntH allowed cells to accumulate Cd²⁺, Zn²⁺, and Co²⁺ as well as Ni²⁺ and Cu²⁺, but to a lesser extent, suggesting that Ni²⁺ and Cu²⁺ represent poor MntH substrates (Makui et al. 2000).

Another study measured the apparent K_M of Mn²⁺ uptake to be 1 μM in E. coli and 100 nM in Salmonella typhimurium. In S. typhimurium, transport of Mn²⁺ was not inhibited by Na⁺, K⁺, Mg²⁺, or Ca²⁺. Contrary to the results described above, Ni²⁺, Cu²⁺, and Zn²⁺ required a 1000-fold excess to inhibit Mn²⁺ uptake, suggesting they are not potential substrates of MntH. However, Cd²⁺ greatly inhibited Mn²⁺ transport (Kehres et al. 2000). Similar observations were seen for EcMntH. Unsurprisingly, though Fe²⁺ could be accumulated by both Ec- and StMntH, transport of Fe²⁺ was significantly inhibited by Mn²⁺. These results corroborate MntH’s preference for Mn²⁺ even though it is capable of transporting other divalent metal ions (Kehres et al. 2000). Consistent with this specificity, regulation of EcMntH is accomplished via the Mn²⁺-responsive transcriptional regulator, MntR, which belongs to the DtxR family, although partial repression by Fe²⁺-Fur was also observed, indicating cross-talk between Mn and Fe homeostasis in this organism (Patzer and Hantke 2001).

The ZupT divalent metal transporter (Fig. 12.2, yellow) from E. coli is also believed to transport Fe²⁺ in addition to Zn²⁺ and other metals (Grass et al. 2002, 2005a). This protein belongs to the zinc-regulated transporter (ZRT), iron-regulated transporter (IRT)-like proteins known as the ZIP family. This family has been identified in animals, plants, protists, and fungi, and these transporters are known to mobilize Fe²⁺, Zn²⁺, Mn²⁺, and Cd²⁺ (Guerinot 2000). ZupT is the first member of this family to be found in the bacterial kingdom (Guerinot 2000; Grass et al. 2002, 2005a). Strains of E. coli defective for all other iron transport systems except for ZupT were able to grow in the absence of metals and in the presence of Mg²⁺, Zn²⁺, Mn²⁺, and Fe³⁺. ⁵⁵Fe uptake assays demonstrated that E. coli strains defective for iron transport but harboring an inducible plasmid for ZupT demonstrated the ability to import Fe²⁺ (Grass et al. 2005a). ZupT was also able to transport Co²⁺ in a strain carrying an additional deletion for CorA, a Co²⁺ transporter. Further demonstrating its substrate promiscuity, ZupT is also believed to transport Mn²⁺, but at a much lower affinity (Grass et al. 2005a). When compared to MntH and Feo (vide infra), the ZupT protein was not as efficient for Fe²⁺ transport, suggesting that ZupT could be a secondary Fe²⁺ transporter in E. coli (Grass et al. 2005a). Despite evidence of broad divalent metal ion substrate specificity by ZupT, this protein remains largely uncharacterized, and more work is needed to understand the molecular details of the metal binding and transport process to determine how ZupT contributes to Fe²⁺ uptake in E. coli.

3.2 ATP-Binding Cassette (ABC) Transporters

The Salmonella iron transporter (sit) operon in Salmonella typhimurium was first identified in 1999 in the centisome 63 pathogenicity island and has high homology to the yfe system (Bearden et al. 1998) from Y. pestis. The sit operon encodes for four proteins: SitA, a putative periplasmic binding protein with homology to YfeA; SitB, the ABC transporter; and SitC and SitD, putative integral membrane permeases with homology to YfeC and YfeD (Fig. 12.2, blue) (Zhou et al. 1999). To test the role of the Sit system in Fe²⁺ transport, StSitABCD was introduced into an enterobactin-deficient strain of E. coli. Expression of the sit operon rescued a growth defect of the E. coli strain in iron-limited media (Zhou et al. 1999). A ΔsitBCD deletion strain was tested in vivo to determine the effect of the Sit system on S. typhimurium virulence; however, no difference in the virulence of the ΔsitBCD deletion strain was detected, which is likely the result of redundant iron transport systems present in S. typhimurium (Zhou et al. 1999). A different study demonstrated that sitABCD is induced in mouse liver when infected intraperitoneally, but induction of sitABCD was much lower in other organs when mice were infected orally. These results suggest that sitABCD is expressed during late-stage infection, after invasion of the intestines (Janakiraman and Slauch 2000). In contrast to previous results, S. typhimurium strains bearing a sitA mutation, which was polar on downstream sit genes, exhibited growth and survival defects in mice (Janakiraman and Slauch 2000). Additionally, the sit operon was found to be controlled by Fe²⁺ and Fur, though another study determined MntR and Mn²⁺ also regulated the sit operon (Zhou et al. 1999; Janakiraman and Slauch 2000; Ikeda et al. 2005). Similar regulation is also observed for the Yfe system of Y. pestis (Bearden et al. 1998; Perry et al. 2012).

Though the Sit system was first identified as an ATP-dependent Fe²⁺ uptake system, it is also able to transport Mn²⁺ with a higher affinity than that of Fe²⁺. Similar to MntH, the apparent affinity of ⁵⁴Mn²⁺ transport by S. typhimurium SitABCD was 0.1 μM and was not affected by pH. In contrast to the association constant, maximal Mn²⁺ transport was pH-dependent and increased at alkaline pH (Kehres et al. 2002), and Fe²⁺ inhibition of Mn²⁺ transport was also observed. Somewhat similarly, transport of ⁵⁵Fe²⁺ by SitABCD occurred in a pH-dependent manner and was inhibited by Mn²⁺ though inhibition was not pH-dependent (Kehres et al. 2002). Cd²⁺, Zn²⁺, Co²⁺, Cu²⁺, Ni²⁺ and Fe³⁺ were also tested for their ability to inhibit Mn²⁺ transport with Cd²⁺ and Zn²⁺ being the most potent inhibitors (Kehres et al. 2002). Whether Fe²⁺ is a natural substrate of the Sit system remains debated and may be dependent upon the conditions under which the Sit system is expressed.

3.3 Oxidase-Dependent Fe²⁺ Transporters (OFeTs)

The elemental ferrous iron uptake (EfeU) protein (also known as YcdN) is a unique Fe²⁺ transporter that has been well characterized but is operative only in select pathogenic species (Fig. 12.2, green). Belonging to the ycdNOB operon, EfeU is homologous to the yeast high-affinity ferric iron permease protein Ftr1p that belongs to the oxidase-dependent iron transporters (OFeTs) (Grosse et al. 2006). These transporters are part of a larger family of proteins known as the iron/lead transporter (ILT) superfamily that transports Fe^2+/3+ and Pb²⁺ (Saier et al. 2006, 2021). Yeast Ftr1p contains an REXXE iron-binding motif, similar to other metal transporters (Stearman et al. 1996; Severance et al. 2004; Grosse et al. 2006), and yeast Ftr1p functions in tandem with a multicopper ferroxidase, Fet3p, and an extracellular reductase, Fre, to transport iron. Fre must first reduce Fe³⁺ to Fe²⁺, at which point Fet3p binds the metal and oxidizes the ion back to Fe³⁺. A hand-off of the metal occurs between Fet3p and Ftr1p (Askwith et al. 1994; Askwith and Kaplan 1997). By comparison, in E. coli, the efeUOB operon (also termed the ycdNOB operon) is polycistronic and is expressed under iron-deplete conditions. Expression of efeUOB is controlled by FUR and is not regulated in response to O₂ (Grosse et al. 2006; Cao et al. 2007). When pH-dependent expression of efeU was determined, efeU displayed higher expression at lower pH (pH = 5.0), under which Fe²⁺ would be stabilized, than at higher pH (pH = 8.0) for both iron-deplete and iron-replete conditions. This pH-dependent regulation is accomplished by cpxAR, a two-component sensor regulator, where CpxA is a histidine kinase localized to the inner membrane and CpxR is an OmpR-like response regulator (Cao et al. 2007). Given EfeU’s homology to yeast Ftr1p, these results could provide insight into the mechanism of iron uptake by EfeU.

EfeU is a 276 amino acid protein predicted to have seven TM helices with a periplasmic N-terminal domain and a cytoplasmic C-terminal domain. REXXE iron binding motifs are located in TM helix 1 and TM helix 4. Additionally, an ≈40 amino acid periplasmic extension is one of two glutamate-rich periplasmic regions located between TM helix 6 and TM helix 7. The TM region is devoid of both negatively charged residues and His residues and, with the exception of the REXXE motifs and some Met residues, appears to be lacking in other metal-binding amino acids (Grosse et al. 2006). To test the important of the REXXE motifs in EfeU, each motif was individually mutated. Regardless of which motif harbored the mutation, an iron uptake deficient strain of E. coli was unable to grow. These results suggest that mutation of the REXXE motifs rendered the organism incapable of binding and transporting iron for growth under experimental conditions. EfeU is also believed to be highly-specific for iron, as EfeU was found to not be involved in the transport of Zn²⁺, Pb²⁺, or Cu²⁺(Grosse et al. 2006), and a preference for Fe²⁺ over Fe³⁺ was also observed, indicating high specificity of EfeU (Cao et al. 2007). EfeU was also tested for in vitro metal transport into proteoliposomes to verify which oxidation state of iron is transported by the protein. Using Phen Green SK as a fluorescence reporter for metal transport into the proteoliposomes, only Fe²⁺ translocation was observed. This transport process is believed to occur via facilitated diffusion, though the energy coupling process has not been fully elucidated (Grosse et al. 2006). It is also possible that additional factors may contribute to Fe²⁺ binding and transport of EfeU, such as the proteins EfeB and EfeO.

EfeB is believed to be a periplasmic paralog of the dye-decolorizing peroxidases. In E. coli, the EfeB protein contains a twin-arginine signal sequence suggesting the protein is translocated across the cytoplasmic membrane to the periplasm via the twin-arginine-translocation (Tat) system. However, some EfeB has also been detected in the cytoplasm. EfeB is known to be a heme b containing-protein regardless of compartmental localization (Sturm et al. 2006). EfeB was confirmed to have peroxidase activity by monitoring H₂O₂-dependent oxidation of guaiacol, a naturally-occurring organic compound; however, no specific substrate of EfeB was identified (Sturm et al. 2006). A later study evaluated the ability of EfeB to degrade heme for iron utilization in E. coli. Soluble fractions of cell lysates containing overexpressed EfeB were able to extract iron from hemin (Létoffé et al. 2009). As no other protoporphyrin metabolites were identified upon iron release from heme by YfeX or EfeB, it was suggested that both proteins have deferrochelation activity and thus leave the protoporphyrin ring intact. It is possible that, in the context of the Efe system, EfeB serves to demetallate heme to provide iron, likely Fe²⁺, to EfeU for transport into the cytoplasm (Létoffé et al. 2009). Both EfeB and its paralog YfeX were subsequently found to be essential for exogenous heme iron acquisition in vivo and strains lacking either yfeX or efeB or both could utilize exogenous heme when complemented with pigA, a heme oxygenase from Pseudomonas aeruginosa, suggesting their functions are related to the release of iron from heme (Létoffé et al. 2009). In the absence of functional heme-uptake systems, overexpression of a functional efeUOB operon allowed for heme acquisition in E. coli.

To further investigate the function of EfeB, the protein was crystallized in complex with heme and the structure was determined to 1.95 Å resolution (PDB ID: 2Y4E). The asymmetric unit consisted of homodimeric EfeB (residues 48–422) in which both molecules were bound to heme (Liu et al. 2011). The crystal structure revealed that EfeB is structurally similar to the de-colorizing peroxidase enzymes and adopts a ferredoxin-like fold. The N-terminal and C-terminal domains of EfeB are connected via a 22-amino acid loop termed the switch loop. This loop helps to stabilize the heme-EfeB interaction solely by interaction with the heme cofactor, potentially to mediate demetallation of the heme cofactor (Liu et al. 2011). While EfeB removes iron from heme in vivo, resulting in the accumulation of protoporphyrin-IX, purified EfeB does not display this enzymatic activity, likely due to the absence of an additional cofactor or cosubstrate. Though EfeB is undeniably involved in the iron transport process, the mechanism of demetallation, and what additional factors may be involved in this process (such as peroxidase activity), remain unknown. Given the role of iron in bacterial pathogenesis, EfeB could serve as an antibacterial target, an endeavor that would be facilitated by further characterization of EfeB and its homologs.

EfeO (YcdO) is a predicted periplasmic protein that has also been found to be similar to dye-decolorizing peroxidases. Similar to EfeB, E. coli EfeO was also predicted to be a Tat-substrate; however, its twin-arginine signal sequence is not conserved amongst other EfeO homologs (Sturm et al. 2006). Though EfeO is not well characterized, EfeO proteins have been grouped into five classes based on their domain architectures: class I, comprising a N-terminal cupredoxin-containing domain (Cup domain) and a C-terminal peptidase-M75 or imelysin-like domain; class II, comprising only a M75 domain (termed EfeM); class III, comprising an unidentified N-terminal domain and a M75 domain; class IV, comprising a class I EfeO fused to EfeU; and class V, comprising only a Cup domain (Rajasekaran et al. 2010). The Cup domain is posited to be involved in electron transfer and also has potential metal binding sites for Cu²⁺ and Fe³⁺. Within the M75 domain is a conserved HXXE motif, a putative metal binding site through which iron could be transferred to EfeU (Rajasekaran et al. 2010), but the validity of this hypothesis remains unclear. More work will be necessary to understand the structure, function, and metal binding properties of EfeO to determine its role in the iron transport process.

3.4 Iron Transporter/More Regions Allowing Vacuolar Colocalization N Protein (IroT/MavN)

Legionella pneumophila is the causative agent of Legionnaires’ disease, and when L. pneumophila infects host cells, it establishes a Legionella-containing vacuole (LCV) associated with the endoplasmic reticulum. Iron is required for the survival and virulence of intravacuolar pathogens, and IroT/MavN is a key transporter that functions at this location (Fig. 12.2, orange) (Portier et al. 2015; Isaac et al. 2015). A substrate of the intracellular multiplication/defect in organelle trafficking (Icm/Dot) type IV secretion system of the LCV, mavN is an iron-regulated gene encoding a 660 amino acid, 75 kDa protein that contributes to Legionella growth and iron transport in both amoeba and macrophages (Isaac et al. 2015; Portier et al. 2015; Christenson et al. 2019). Additionally, L. pneumophila bearing a non-functional MavN exhibited growth defects on iron-restricted agar and acquired Fe²⁺ at levels lower than wild type (WT) strains (Portier et al. 2015). These results indicate that MavN is important for iron acquisition to support the growth and the survival of this pathogen living within a variety of different environments.

Topologically, MavN is predicted to have 8 TM helices with both the N- and C-termini located in the cytoplasm. MavN also contains four predicted cytoplasmic loops and three predicted loops in the LCV lumen. This topology was verified by the transmembrane-substituted cysteine accessibility method (Christenson et al. 2019; Isaac et al. 2015). Like other iron-binding proteins, MavN also contains EXXE motifs and nine of these motifs can be found within the protein. An EXXE motif in a 70-amino acid loop (denoted at loop 7) mapped to the LCV lumen and conserved amongst all Legionellaceae species was the only one of these motifs determined to be wholly essential for L. pneumophila growth, as a plasmid carrying this mutation could not complement a ΔmavN growth defect (Isaac et al. 2015; Christenson et al. 2019). This loop also contains eight His residues, three of which were found to be important for MavN function in a triple mutant (Isaac et al. 2015). Only one His residue is predicted to be localized to the TM region along TM helix 6, and mutation of this residue exhibits a growth defect. Additionally, of the four Met residues in MavN, only one (along TM helix 3) led to a growth defect in vivo (Christenson et al. 2019). These in vivo results were subsequently used to test an in vitro transport model of MavN.

For in vitro analysis of MavN, the protein was heterologously expressed and purified from the methylotrophic yeast Pichia pastoris. Solubilization and purification in detergents C₁₂E₆ and C₁₂E₈ and yielded predominately dimeric protein (Christenson et al. 2019). To test substrate binding, Co²⁺ and Ni²⁺ were tested for their ability to bind to MavN as surrogates for Fe²⁺ under oxic conditions, and both substrates bound with modest affinity (K_d ca. μM) (Christenson et al. 2019). MavN was then reconstituted into a liposome to test for metal ion transport using calcein, a dye whose fluorescence is quenched by metal binding. Transport of several metal ions into proteoliposomes were tested and a K_M for Fe²⁺ transport was measured to be 16 μM. Though transport of Mn²⁺, Co²⁺, and Zn²⁺ into the proteoliposomes was also observed, transport of Cu⁺, Cu²⁺, and Ni²⁺ was not (Christenson et al. 2019). To determine whether Mn²⁺ and Zn²⁺ could serve as substrates for MavN, these metals were provided in excess to a ΔmavN strain growing in bone marrow deficient macrophages, in addition to Fe²⁺. In these experiments, both Mn²⁺ and Zn²⁺ stimulated growth of the ΔmavN strain, suggesting substrates other than Fe²⁺ could be translocated by MavN in vivo (Christenson et al. 2019). In vitro studies of Fe²⁺ transport into proteoliposomes with variant MavNs showed an importance for His and Met residues in Fe²⁺ translocation, while Cys variants only had modest effect on Fe²⁺ transport (Christenson et al. 2019).

Another study published shortly after also reconstituted MavN into liposomes for metal transport studies. In this study, MavN was heterologously expressed in E. coli, extracted from membranes with n-tetradecylphosphocholine (Fos-choline-14) and then purified in 7-cyclohexyl-1-heptyl-β-d-maltoside (Cymal-7), which resulted in chiefly monomeric protein (Abeyrathna et al. 2019). MavN was then incorporated into liposomes containing Fluozin-3, a turn-on fluorescent probe, to monitor transport activity. Transport of Fe²⁺ was monitored in an anoxic environment and similar K_d values (μM) were observed (Abeyrathna et al. 2019). However, contrary to the previous study, transport of Fe²⁺ was highly specific, and Co²⁺, Ni²⁺, and Zn²⁺ provided <5% of the Fe²⁺ response. As a control, Fe³⁺ was also tested for transport, and was not taken up by MavN. These results suggest that MavN could be highly specific for Fe²⁺ (Abeyrathna et al. 2019). Because of the acidic pH of the LCV, it was suggested that Fe²⁺ transport could be coupled to proton counter-transport. To test this, a pH indicator, pyranine, was incorporated into the proteoliposomes. Subsequently, as Fe²⁺ uptake occurred, the pH inside the vesicles increased, suggesting that MavN could function as an Fe²⁺/H⁺ antiporter or as a secondary active transporter (Abeyrathna et al. 2019). To determine which residues may contribute to Fe²⁺ uptake, a series of variant proteins were constructed and incorporated into liposomes. A variant protein containing mutations to Glu and His residues in TM helices 2 and 3 and Asp and Cys residues in TM helix 8 was constructed and tested for transport. This variant maintained the same V_max as the WT protein; however, K_M was higher (>25 μM) (Abeyrathna et al. 2019). A different variant of Glu, Cys, and His residues at the TM-LCV interface abolished Fe²⁺ uptake, indicating that these residues could serve to release Fe²⁺ into the vacuole (Abeyrathna et al. 2019). Taken together, these in vitro assays to assess metal transport will be useful models for determining real-time metal transport of other transporters, especially for determining residues involved in the transport process. If used in conjunction with other biochemical and biophysical techniques, transport mechanisms could be determined, which could aid in the design of molecules that inhibit the transport process.

While the systems described above are known or are purported to be Fe²⁺ transport systems, they remain largely uncharacterized at both the cellular and the atomic levels. Furthermore, some of these transporters display a broad substrate specificity and they may have a relatively weak affinity for Fe²⁺, at least under the tested experimental conditions. Additionally, some transporters such as ZupT do not appear to be broadly distributed amongst bacteria, though it is possible that homologs have yet to be identified. The Efe system, though specific for Fe²⁺, is only distributed in some pathogenic bacteria. To our knowledge, the only widespread, prokaryotic, Fe²⁺-specific transport system to be identified and characterized in several organisms is the ferrous iron transport (Feo) system, which is described in detail below.

3.5 Ferrous Iron Transport by the Feo System

The Feo system is the predominant prokaryotic Fe²⁺ transport system (Fig. 12.2, purple). The feo operon, encoding for the Feo system, was first discovered in 1987 by isolating an E. coli strain defective for Fe²⁺ uptake (Hantke 1987). At that time, five E. coli Fe³⁺ transport systems for ferric siderophores had been identified, but no system had been identified for Fe²⁺ transport. The locus responsible for Fe²⁺ transport was then termed feo, but its position within the E. coli genome remained unknown at the time. Feo was subsequently identified to be a high affinity transporter, as cellular uptake studies revealed an apparent K_M of Fe²⁺ to be ≈0.5 μM, and in 1993, the feoA and feoB genes were cloned and their sequences were published (Kammler et al. 1993). The Feo system also appears to be under regulation by the fumarate and nitrate reduction regulator (FNR) as FNR binding sites were found near the genes encoding for FeoA and FeoB (Fig. 12.4). FNR is a cAMP-like transcriptional regulator that senses oxygen using a [4Fe-4S] cluster and binds to DNA using a helix-turn-helix domain. The [4Fe-4S] cluster is degraded in the presence of oxygen, and this cluster change controls DNA binding and alters a metabolic “switch” that converts bacterial metabolism from anaerobic to aerobic respiration (Spiro and Guest 1990; Unden et al. 2002; Crack et al. 2008). Additionally, a FUR binding site (ferric uptake regulator, one of the master regulators of general prokaryotic iron metabolism) was also discovered downstream from the FNR binding site (Fig. 12.4) (Kammler et al. 1993). When cellular iron is low, the Fe²⁺-FUR complex dissociates and allows for transcription of iron-responsive genes (Escolar et al. 1999; Troxell and Hassan 2013; Fillat 2014). Unsurprisingly, a fur mutant was observed to accumulate Fe²⁺ more rapidly than WT, indicating feo to be under FUR regulation. Thus, Feo-mediated Fe²⁺ transport is highly regulated, but whether additional levels of regulation exist for this system remains unclear.

Fig. 12.4

The most common arrangements of the feo operon. (a) The operon encoding for FeoA, FeoB, and FeoC was first discovered in E. coli in 1987. However, the FeoC protein is poorly conserved (found in ≈13% of bacteria), which is denoted by a dashed line. These genes are downstream of the binding sites for the FNR and FUR regulators. (b) The most common operon arrangement encodes for only FeoA and FeoB. (c) Though rare, some feo operons encode for an FeoA-FeoB fusion, such as in Bacteroides fragilis. (d) Some feo operons encode for FeoB alone, demonstrating the essential nature of this gene. Figure created with BioRender

As more genomic data have become available, it has been proposed that Feo represents an ancient Fe²⁺ transport system (Hantke 2003), which is supported by several lines of evidence. First, as previously described, it is clear that Fe²⁺ was the predominant species of iron present on Earth as bacteria first evolved, undoubtedly necessitating an Fe²⁺ uptake system. Second, several cyanobacterial descendants (among the first unicellular organisms to exist on the planet) retain an FeoB protein within their genomes, suggesting that this system has played a major role during their evolution to necessitate such prolonged genetic retention. In fact, this conservation suggests that despite the switch from an anoxic to a more oxygen-rich atmosphere, Fe²⁺ transport was still necessary for the survival of many organisms. FeoB may have co-evolved alongside organisms to supply metabolic feedstocks of Fe²⁺, a strategy that is undoubtedly still operative on modern Earth. Underscoring this point is the major conservation and distribution of the feo operon across present-day bacterial species. As of 2003, it was estimated that ≈50% of all bacterial genomes that had been sequenced at that time contained an feoB-like gene (Hantke 2003), emphasizing the dominance of Feo in prokaryotic Fe²⁺ acquisition.

The organization of the tripartite feoA/feoB/feoC operon found in E. coli is considered to be canonical, as it was the first to be discovered (Fig. 12.4a) (Hantke 2003, 1987; Kammler et al. 1993). However, there are several other arrangements of the feo operon, and it is now clear that a tripartite system is unlikely to be the dominant arrangement in most sequenced prokaryotic genomes. The tripartite operon arrangement is predominantly found in the γ-proteobacteria class but can also be found in terrabacteria (Cartron et al. 2006; Lau et al. 2016; Sestok et al. 2018, 2021). The FeoA protein is most commonly found alongside FeoB and some operons are predicted to encode for multiple FeoA proteins (Fig. 12.4b) (Lau et al. 2016; Cartron et al. 2006; Sestok et al. 2018). A recent search of the FeoA protein (IPR007167) using the InterPro Database (accessed February 2021) revealed that ≈3% of all FeoA proteins exist as fusions of either two or three FeoA proteins (Sestok et al. 2021). FeoA can also exist as a naturally-occurring fusion to the N-terminal domain of FeoB (Fig. 12.4c) (Sestok et al. 2018; Lau et al. 2016; Veeranagouda et al. 2014; Rocha et al. 2019; Dashper et al. 2005); these fusions are rare, found predominantly in the Bacteroidia and Clostridia classes, and account for only ≈8% of all FeoA-domain containing proteins in the InterPro Database (Sestok et al. 2021). Finally, some bacterial genomes encode for only FeoB (Fig. 12.4d).

Regardless of the arrangement of the organism’s operon, it is clear that FeoB is the chief component of the Feo system based on its conservation (Fig. 12.4). FeoB is a complex, polytopic membrane protein thought to resemble a covalent fusion of a G-protein to that of a G-protein-coupled receptor (GPCR). Guanine nucleotide-binding proteins, termed G-proteins, are a family of intracellular proteins and are associated with transmitting signals from external stimuli to the cell’s interior. Each protein has two states, termed active and inactive, and for this reason G-proteins have been likened to molecular versions of switches. Consistent with this description, FeoB is typically composed of three main domains with their residues numbered relative to E. coli K-12 FeoB (Uniprot ID P33650): the G-protein domain that binds and hydrolyzes GTP (residues 1–170), the guanine dissociation inhibitor (GDI) domain (residues 171–276) that increases GDP affinity, and the TM region (residues 277–773) that has an ill-defined function. The G-protein domain and the GDI domain comprise the full N-terminal soluble domain of FeoB, which is termed NFeoB. To date, the majority of published literature on the Feo system has focused solely on NFeoB (vide infra), likely due to its more tractable, soluble form and its intriguing similarity to well-characterized G-proteins.

G-proteins play important roles in signal transduction in eukaryotic cells, but their roles in bacterial physiology are not well understood. In eukaryotes, G-proteins are numerous and as such are responsible for an array of cellular processes such as nuclear import and export, vesicle formation, exocytosis, regulation of the cytoskeleton, and cellular differentiation (Bos et al. 2007). In contrast, bacterial G-proteins may be involved in ribosome biogenesis, tRNA modification, cell cycle progression, or DNA replication. G-proteins cycle between on and off states, dependent on the identity of the nucleotide bound to the protein to carry out their molecular functions (Vetter and Wittinghofer 2001). GDP dissociation from a G-protein results in activation, which is followed by GTP binding and hydrolysis, returning the protein to its inactive form (Vetter and Wittinghofer 2001). Interestingly, bacterial G-proteins are structurally similar to eukaryotic G-proteins, but their nucleotide binding affinities are several orders of magnitude lower. Because eukaryotic GTPases display high nucleotide binding affinities and intrinsically slow GTP hydrolysis and GDP dissociation, G-activating proteins (GAPs) or guanine exchange factors (GEFs) are needed either to increase the rate of GTP hydrolysis or to facilitate the release of GDP (Vetter and Wittinghofer 2001). Regions within the G-protein domain are also important for how guanine nucleotides are recognized, and how these proteins interact with downstream effectors. Conformational changes are observed in these regions (termed switch I and switch II), dependent upon the nucleotide-state of the G-protein (Bos et al. 2007; Vetter and Wittinghofer 2001). Many of these aspects have been explored with regards to NFeoB function, and an abbreviated synopsis is presented below.

3.5.1 Composition and Structure of NFeoB

E. coli FeoB (EcFeoB) was demonstrated to contain a G-protein domain in 2002, representing the first example of a G protein tethered to a prokaryotic membrane protein (Marlovits et al. 2002). At that time, >95 species containing FeoB homologs had been identified. There was (and still is) no predicted homology between the TM domain of FeoB and that of other transmembrane proteins (Marlovits et al. 2002). Using multiple sequence alignments, four of the five G-protein motifs were identified in EcFeoB that are responsible for binding and hydrolyzing GTP (Table 12.1 and Fig. 12.5) (residues numbered based on E. coli K-12 FeoB; Uniprot ID P33650): G1 (residues 10–17; Fig. 12.5, purple), G2 (residue 37; Fig. 12.5, yellow), G3 (residues 56–59; Fig. 12.5, salmon) and G4 (residues 120–123; Fig. 12.5 blue). At the time, the G5 motif (Fig. 12.5, gray) could not initially be identified because of poor sequence conservation (Marlovits et al. 2002).

Table 12.1 Consensus sequences for G-protein motifs in NFeoB. The G1–G4 motifs are conserved across bacterial taxonomic groups and were initially identified through multiple sequence alignments. The G5 motif was identified through structural alignments. All amino acid numberings are based on EcFeoB

Fig. 12.5

X-ray crystal structure of E. coli NFeoB (PDB ID: 3I8S), comprising the GDI and G-protein domains. Five G-protein motifs are found in NFeoB and are responsible for the recognition and the binding of guanine nucleotides. The Pro-rich hydrophobic sequence in NFeoB is located along a β strand adjacent to the G5 motif and is believed to be the site of FeoA-FeoB interactions. ‘N’ and ‘C’ represent the N- and C-termini respectively. Figure created with BioRender

In 2008, the linker region between the G-protein domain and the TM region in FeoB was identified as a GDI domain. Initially, a weakly-conserved LXXXE motif in the TM linker region identified by sequence alignments was believed to resemble a binding site for a GEF. An E210A variant was generated in both EcNFeoB and full-length EcFeoB. The EcNFeoB^E210A variant exhibited a decreased 3′-O-(N-Methyl-anthraniloyl)-GTP (mant-GTP) affinity by approximately two-fold, but the mant-GDP affinity decreased by approximately four-fold. The authors attributed the function of the linker region between the G-protein domain and the TM region to that of a GDI domain instead of a GEF. The EcFeoB^E210A variant was not functional in vivo, suggesting the GDI domain could play an important role in FeoB function (Eng et al. 2008). Subsequently, each domain was tested for the ability to affect nucleotide binding affinities. The presence of the GDI domain increased the affinity of mant-GDP by 13-fold, and the presence of the TM region further increased the affinity for mant-GDP by 400-fold. The binding affinity for mant-GTP by the G-protein domain and NFeoB were similar; however, the full-length EcFeoB exhibited much stronger binding of mant-GTP, further demonstrating that this region functions as a GDI domain (Eng et al. 2008). Interestingly, multiple sequence alignments of the NFeoB domains from E. coli, Salmonella typhimurium, Vibrio cholerae, and Helicobacter pylori showed that the linker region varies in sequence and length, which could be a functional variation among species for this domain. Subsequent structural analyses then provided further insight into this intriguing domain.

Some of the first structures of the G-protein domain of FeoB (residues 1–184) were determined in 2009 from the archaeon Methanococcus jannaschii in the apo (PDB ID: 2WJH), GDP-bound (PDB ID: 2WJG), and GMP-PNP bound (PDB ID: 2WJI) forms (Koster et al. 2009a, b). M. jannaschii NFeoB (MjNFeoB) crystallized as a homodimer in the presence and absence of nucleotides, in which the nucleotide binding pockets form the dimeric interface. The overall structure of MjNFeoB revealed six α-helices and a core β-sheet composed of seven β-strands. However, major structural differences exist in the G-domain dependent upon the nucleotide state of the protein. In the GTP-bound form, the Switch I loop points away from the nucleotide binding site and the β-sheet in the GTP-bound form lacks one β-strand (residues 32–36), which could not be seen in the electron density. This suggests that the Switch I loop, located between the G1 and G2 motif, may be more flexible in the GTP-bound form. Additionally, structural alignments of MjNFeoB with eukaryotic GTPases revealed the G5 motif, not previously identified through sequence alignments, to be located at position 145–148 (SAAK) (Koster et al. 2009b).

The Streptococcus thermophilus NFeoB was crystallized in the presence of GDP⋅AlF₄⁻, a transition-state analog that locks the G-protein in its active state, providing insight into how GTP hydrolysis is initiated (Ash et al. 2011). StNFeoB was co-crystallized with GDP⋅AlF₄⁻, one K⁺ ion, and one Mg²⁺ ion (PDB ID: 3SS8) (Ash et al. 2011). This transition-state structure of StNFeoB captured the Switch II region pointing away from the nucleotide binding site, in contrast with the Switch I region that caps the nucleotide binding site, as observed in the previous mant-GMP-PNP bound structures (Ash et al. 2011). The K⁺ ion is bound in the nucleotide binding site with coordination by Gly29, Trp31, and Asn11, similar to its position in 3LX5, while the Mg²⁺ ion, ≈5 Å away from the K⁺, contacts the planar AlF₄⁻ molecule that mimics the γ-phosphate of the nucleotide. Additional AlF₄⁻ interactions include an oxygen atom from the β-phosphate, backbone amides from Gly33, Val34, and Thr35 in the Switch I region, and a water molecule that performs nucleophilic attack (Ash et al. 2011). Importantly, the attacking water appears to be solvent exposed in this structure, which could allow the positioning of another catalytic residue, possibly from a different domain in FeoB or another partner protein (Ash et al. 2011).

In 2013, the X-ray crystal structure of NFeoB from Gallionella capsiferriformans was published in both the apo and the GDP-bound states. G. capsiferriformans provides a unique look into the structure of the cytoplasmic domain of FeoB, as GcFeoB lacks a GDI domain while the G-protein domain is structurally analogous to that of EcNFeoB (PDB ID: 3HYT) (Deshpande et al. 2013). While GcNFeoB was chiefly monomeric in solution, the protein crystallized as a domain-swapped dimer (Deshpande et al. 2013). GcNFeoB also contained the five G-protein motifs but exhibited a highly disordered G5 motif in both the apo structure and the GDP-bound structure (Deshpande et al. 2013). In agreement with the crystal structure of S. thermophilus NFeoB, the Switch I region was oriented away from the GDP binding pocket, supporting the hypothesis that the Switch I region is necessary for nucleotide binding and release (Deshpande et al. 2013).

Another crystal structure of NFeoB from E. coli BL21 was solved in 2016 (Hagelueken et al. 2016). After removal of the nucleotide, which typically co-purified with the protein, EcNFeoB crystallized as a trimer. This structure was similar to other apo NFeoB structures with the switch I region pointing away from the nucleotide binding site (suggested to be in the “open” conformation) (Hagelueken et al. 2016).

Numerous X-ray crystal structures of variant NFeoBs from several organisms have since been solved. However, these structures have failed to answer questions about how the Switch I and Switch II regions may be involved in regulating nucleotide binding and release, Fe²⁺ transport, and potential protein-protein interactions. Likewise, structural changes that might occur in the GDI domain upon metal binding and release have not been determined because FeoB lacks the TM region, further underscoring the importance of undertaking structural studies of full-length FeoB.

3.5.2 NFeoB Function and GTPase/NTPase Activity

Early seminal results revealed the ability of NFeoB to bind guanine nucleotides, and experiments have importantly linked the activity of the G-protein domain to Fe²⁺ uptake. Stopped-flow studies of EcNFeoB with non-hydrolyzable guanine and adenine nucleotide analogs, mant-5′-Guanylyl-imidodiphosphate (mant-GMP-PNP) and mant-5′-Adenylyl-imidodiphosphate (mant-AMP-PNP), revealed specificity towards guanine nucleotides (Marlovits et al. 2002). At 20 °C, the K_d of affinity-tagged EcNFeoB for mant-GMP-PNP was determined to be ≈4 μM (based on slow association but fast release kinetics), similar to Era but three orders of magnitude lower than p-21 Ras (Marlovits et al. 2002). However, a follow-up study of EcNFeoB measured a K_d for mant-GTP to be 12 μM, a threefold increase over what was previously observed for mant-GMP-PMP, suggesting the non-hydrolyzable analog may not be a good proxy for GTP binding (Eng et al. 2008). GTP hydrolysis of EcNFeoB was markedly slow at ≈0.0015 s⁻¹ and hydrolysis of ATP was not observed (Marlovits et al. 2002). Initially, two potential sites existed for the G4 motif, residues 91–94 or residues 120–123. To determine which site contained the G4 motif, two variants (D94N and D123N) were tested for their ability to bind mant-GMP-PNP in stopped-flow experiments and only the D94N variant could bind mant-GMP-PNP, confirming the G4 motif is located at positions 120–123 (Marlovits et al. 2002). Because the D123N variant failed to recognize mant-GMP-PNP, it was suggested that D123 could impact Fe²⁺ transport in the full-length protein. In an feoB deletion strain deficient for Fe²⁺ uptake, full-length FeoB was able to rescue Fe²⁺ uptake whereas a plasmid carrying a D123N variant was not. These results thus demonstrated a link between GTP binding, GTP hydrolysis, and Fe²⁺ transport (Marlovits et al. 2002).

Following these results, biochemical and structural experiments have revealed the contributions of some G motifs (and G5 in particular) to NFeoB-catalyzed GTP hydrolysis. Eight EcNFeoB variants located in the G motifs and switch regions, were examined for their effects on nucleotide binding and feoB function in vivo. All variants exhibited at least a two-fold decrease in GDP affinity, while only a T37A variant drastically decreased GTP affinity and could not restore Fe²⁺ uptake in a feoB-deficient strain (Eng et al. 2008). The N32A and D73A variants were able to hydrolyze GTP, but could not restore Fe²⁺ transport in a feoB-deficient strain (Eng et al. 2008). Additionally, the P12G, T60Q, and Y61A variants restored in vivo function of FeoB while the Y61E and P58A variants did not (Eng et al. 2008). Another study on EcNFeoB investigated changes in GDP release as a result of differences in sequence composition in the G5 motif (Álvarez-Fraga et al. 2018). Structural changes in the G5 motif are common in GTPases, but the NFeoB G5 motif seems to have poor sequence conservation and consists of six amino acids (V149–G154). In several structures of NFeoB, the loop formed by the G5 motif interacts with nucleotides through a mixture of H-bonding and hydrophobic interactions between the polypeptide and the nucleotide base (Álvarez-Fraga et al. 2018). This interaction must be important, as this structural motif is conserved despite low sequence conservation. An alanine in the second position in the G5 motif is typically the most conserved residue, and mutation of this analogous residue in eukaryotic GTPases results in severe health complications (GDP is released so fast the protein is always “on”) (Álvarez-Fraga et al. 2018). Interestingly, E. coli FeoB does not conserve an alanine residue, but rather a serine residue in this position. For WT EcNFeoB, GTPase activity was measured at 0.40 min⁻¹ in the presence of K⁺, a sevenfold increase than in the presence of Na⁺ (Álvarez-Fraga et al. 2018). Subsequently, each residue was separately mutated to an alanine. All variants, with the exception of a S150A variant, displayed lower GTPase activity. The S150A variant’s GTPase activity was 1.5-fold faster than the WT protein and exhibited a seven-fold slower release rate of mant-GDP (Guilfoyle et al. 2014). Using ITC, the S150A variant bound GDP five-fold higher than the WT, which correlates well with an observed slower GDP release (Álvarez-Fraga et al. 2018). Interestingly, differences in nucleotide binding and release are attributed to alterations in the hydrogen bonding network to the nucleotide base when serine is present versus alanine, destabilizing the nucleotide base. While the sequence composition of the G5 loop is clearly important for nucleotide binding and release, it remains unknown whether GDP release is a result of loop movement or if loop movement is a result of GDP release (Álvarez-Fraga et al. 2018).

Despite structural homology to eukaryotic G-proteins, recent publications have suggested that some FeoB proteins may be NTPases rather than strict GTPases. The first study to explore this idea found that VcNFeoB could hydrolyze both GTP and ATP, with a preference for GTP over ATP. Interestingly, VcNFeoB was also able to hydrolyze inosine triphosphate (ITP), but saturating conditions were never achieved. In contrast, EcNFeoB hydrolyzed GTP and ITP but did not hydrolyze ATP (Shin et al. 2019). Two variants in the G5 motif (S148T and N150T), which are responsible for interactions with the nucleotide base, displayed decreased ATPase activity but were able to transport Fe²⁺ in vivo, suggesting the variant VcNFeoBs were functional GTPases but not functional NTPases (Shin et al. 2019). Helicobacter pylori NFeoB was also tested for NTPase activity in this study and was able to hydrolyze GTP, ITP, and ATP, thus leading to the classification of HpNFeoB as an NTPase (Shin et al. 2019). VcFeoA and VcFeoC were also tested for their effects on GTPase and ATPase activity of VcNFeoB. When VcFeoA and VcNFeoB were mixed at a 1:1 ratio in the presence of 650 μM ATP, ATP hydrolysis was reduced by ≈60%, and the addition of VcFeoC at a 2:1:2 ratio did not have any further effect on ATPase activity. Conversely, VcFeoA and VcFeoC stimulated GTPase activity of VcNFeoB at a 2:1:2 ratio. These results suggest that, at least in V. cholerae, FeoA and FeoC could function to regulate nucleotide hydrolysis activity (Shin et al. 2019). These observations raise interesting questions about the roles of FeoA and FeoC and whether their functions differ depending on whether their cognate FeoBs are strict GTPases or promiscuous NTPases.

A follow-up study further characterized NFeoB proteins from different organisms and suggested a sequence dependence on GTPase/NTPase classification (Shin et al. 2020). The authors found that NFeoBs from E. coli, Salmonella enterica serovar Typhimurium, and Pseudomonas aeruginosa (all Gram-negative pathogens) were only capable of hydrolyzing GTP, whereas NFeoBs from Streptococcus mutans, Staphylococcus aureus, and Bacillus cereus (all Gram-positive pathogens) were able to utilize both GTP and ATP (Shin et al. 2020). Through the use of multiple sequence alignments, the presence of a serine or alanine at position 150 (in EcNFeoB) was used to predict whether a NFeoB protein would be a GTPase or an NTPase, respectively. Phylogenetic analysis showed that NFeoB proteins with GTPase activity and the conserved Ser appear to be clustered together and predominantly present in γ-proteobacteria, with the exception of V. cholerae and H. pylori, while those with NTPase activity and the conserved Ala are broadly distributed in different phyla (Shin et al. 2020). All WT NFeoBs displayed increased GTP and ATP hydrolysis with increasing temperature. Interestingly, BcNFeoB displayed a preference for ATP at lower temperatures and GTP at higher temperatures (Shin et al. 2020). Why some FeoB proteins would potentially favor dual nucleotide specificity and some would favor GTP as a substrate remains unknown, and further studies investigating nucleotide binding and hydrolysis in full-length FeoBs will be necessary to determine dual nucleotide specificity.

While several NFeoBs have been structurally characterized throughout the GTP-hydrolysis cycle, there has been much debate regarding the functional oligomeric state of NFeoB (and thus full-length FeoB) in vivo. Q-band pulsed electron-electron double resonance (PELDOR) spectroscopy was used to characterize conformational, dynamical, and oligomeric changes in EcNFeoB in the presence and absence of GTP. These data demonstrated that the largest structural changes in NFeoB during GTP binding occur in the switch I region (Hagelueken et al. 2016). A doubly MTSSL spin labelled but apo (nucleotide-free) EcNFeoB was found to adopt an “open” conformation in solution (≈40 Å label distance), similar to that seen in the crystal structure of EcNFeoB. The addition of a 20-fold molar excess of GTP or GDP decreased the distance distribution of the spin labels on the protein to ≈20 Å, which was attributed to the “closed” state of the protein in which the switch I region moves to cover the GTP binding pocket, similar to the PELDOR observations (Hagelueken et al. 2016). This experiment was repeated with a 100-fold molar excess of GMP-PNP, which induced a conformational change from the “open” state to the “closed” states; however, this shift only corresponded to ≈5% of EcNFeoB molecules in the “closed” state (Hagelueken et al. 2016). While the presence of K⁺ did stimulate GTP hydrolysis activity of EcNFeoB, the presence of K⁺ did not affect the conformation of the switch I region (Hagelueken et al. 2016). A K1 spin label was then incorporated into EcNFeoB to investigate oligomeric states present in solution. At low concentrations (25 μM), EcNFeoB was found to be monomeric in solution, but ≈5% of EcNFeoB molecules formed higher-order oligomers at much higher protein concentrations (≈500 μM). While the data suggest EcNFeoB monomers could interact at high concentrations, the authors noted that these interactions do not appear to be specific as the distance distribution observed was broad (Hagelueken et al. 2016). Thus, the exact oligomerization of the soluble NFeoB domain remains an open debate.

3.5.3 Full-Length FeoB

An intellectually curious aspect of FeoB is its unique combination of a G-protein covalently attached to a large TM region, resembling a hybrid of strategies commonplace in nature. Typically, small G-proteins cycle between active (GTP-bound) and inactive (GDP-bound) conformations depending on the nucleotide status of the protein and the cell (Bos et al. 2007). To assist localization of G-proteins to the membrane, their C- and/or N-termini may be prenylated, myristoylated, or acetylated (Bos et al. 2007). GDI domain-lipid tail interactions allow for G-proteins to be removed from the membrane and provide another level of control over G-proteins, in addition to GAPs and GEFs. GDI domain regulation in G-proteins is common in Rho and Rab GTPases (Bos et al. 2007). In the case of FeoB, the G-protein domain is uniquely tethered to its TM region by a GDI domain. It could be possible that interactions between the GDI domain and the TM region provide control over nucleotide binding at the G-protein domain, nucleotide hydrolysis, and even Fe²⁺ transport. However, structural and functional studies on FeoB are scarce, representing a major hole in our understanding of this component of the Feo system.

While no structure of an intact FeoB exists, the first model to appear was that of Pseudomonas aeruginosa FeoB (PaFeoB) (Fig. 12.6). In this model, the predicted structure of the N-terminal domain was based on homology to the crystal structure of EcNFeoB (crystallized as a trimer in the presence of mant-GTP; PDB ID: 3HYT) (Seyedmohammad et al. 2016). The TM region of FeoB was then modelled based on sequence similarity to an archaeal glutamate transporter (crystallized as a trimer; PDB ID: 1XFH). The linker region responsible for joining the soluble PaNFeoB domain with the TM region was modelled based on sequence identity to the EDH2 ATPase (crystallized as a dimer; PDB ID: 2QPT), and the C-terminal domain of PaFeoB was modelled based on sequence identity to dihydrodipicolinate reductase (crystallized as a tetramer; PDB ID: 1DIH) (Seyedmohammad et al. 2016). Despite these oligomeric discrepancies, the entirety of the PaFeoB polypeptide was modelled as a homotrimer, and two conserved cysteine residues in PaFeoB were suggested to be involved in the binding and transport of Fe²⁺ (Fig. 12.6). Blue native PAGE and atomic force microscopy of purified PaFeoB in C₁₂E₈ revealed three oligomeric states to be present: monomer, trimer, and hexamer. Based on the available data, monomeric PaFeoB seems to be the predominant oligomeric state followed by trimeric PaFeoB, with hexameric PaFeoB being the least prevalent (Seyedmohammad et al. 2016). A putative metal-binding residue Cys⁴²⁹, topographically modeled to be in TM helix 4, is predicted to form a ring within the homotrimer model. With an estimated diameter of ≈5 Å, this pore is hypothesized to be in the “open” state of the protein, which would be capable of transporting Fe²⁺ (Seyedmohammad et al. 2016). To probe this hypothesis further, GTPase activity of WT and variant PaFeoBs was measured and tested for stimulation by Fe²⁺. Maximal stimulation of GTPase activity occurred in the presence of 1–1.5 mM Fe²⁺. Furthermore, a C429S variant had no effect on GTPase activity while a C675S variant, topographically modeled to be in TM helix 7 and predicted to be exposed to the periplasmic space, significantly reduced GTPase activity. The authors thus rationalized that Cys⁶⁷⁵ could serve as an Fe²⁺ sensor (Seyedmohammad et al. 2016). While this model provides one starting place for biochemical experiments to probe the mechanism of FeoB-mediated Fe²⁺ transport, another model exists that suggests a different transport mechanism could be possible.

Fig. 12.6

Models of FeoB-mediated Fe²⁺ transport. (a) Our laboratory has proposed that FeoB utilizes a Met-lined channel (yellow) to bind and to transport ferrous iron across the bacterial membrane (Sestok et al. 2018) based on a de novo calculated model of EcFeoB (Ovchinnikov et al. 2017). (b) Others have proposed that FeoB utilizes an oligomerically-induced Cys-lined pore (pink) to bind and to transport Fe²⁺ across the bacterial membrane based on a model of PaFeoB that has modest sequence homology to several functionally different bacterial transporters (Seyedmohammad et al. 2016)

A second model, this time of E. coli FeoB, was published in 2017 (Fig. 12.6) (Ovchinnikov et al. 2017). There are no true homologs of full-length FeoB and, because of its large size, accurate de novo models have been difficult to generate. Unlike the PaFeoB model, which was created by using proteins with modest sequence similarity to PaFeoB, the EcFeoB model was generated by determining residue-residue contacts with metagenome sequence data. This combination of evolutionary data and structural prediction has resulted in the generation of hundreds of structures of proteins from families without previously known structures (Ovchinnikov et al. 2017). Analysis of the de novo EcFeoB, a monomeric structure, reveals within FeoB the presence of a Met-lined channel that could serve as a translocation pathway for Fe²⁺ (Fig. 12.6) (Sestok et al. 2018). Additionally, the Cys residue predicted to be located at the periplasmic face and involved in the formation of a Cys-lined pore in the PaFeoB model (Cys⁶⁷⁵) is located within the TM region in the EcFeoB model (Cys⁶⁷⁷) (Sestok et al. 2018). However, despite the presence of two FeoB models, neither has been validated, and a full-length structure of FeoB is still unrealized, likely due to the difficulty of working with the intact membrane protein.

Despite the lack of atomic-level details of FeoB, at least a few publications have aimed to broaden our understanding of FeoB by addressing the challenge of large-scale expression and purification of the intact protein. In 2014, full-length PaFeoB and a D123N variant were the first particulate FeoB constructs to be solubilized in n-dodecyl-β-d-maltoside (DDM) micelles, to be purified, and to be characterized functionally in vitro. Unstimulated GTP hydrolysis was slow (≈0.0035 s⁻¹), in agreement with the slow hydrolysis rates measured for soluble NFeoB. GTP hydrolysis by the D123N variant (a key G4 motif) was modestly altered to ≈0.0012 s⁻¹, demonstrating the importance of this residue in the full-length protein (Seyedmohammad et al. 2014). Functional reconstitution of PaFeoB into inside-out vesicles was also performed, and the GTP hydrolysis was modestly increased to ≈0.0046 s⁻¹ in the absence of any metal or additional proteins (Seyedmohammad et al. 2014). In 2015, full-length EcFeoB and N-terminal EcNFeoB were both expressed and purified for spin-labelling EPR studies. An unnatural amino acid, para-acetylphenylalanine (pAcF), was incorporated into full-length EcFeoB at Lys¹²⁷ for this purpose (Hagelueken et al. 2015). Labeled EcFeoB was subsequently solubilized and purified in DDM. The dominant sizing profile of purified EcFeoB corresponded to a molecular weight of ≈480 kDa, while EPR experiments used rotational constants of the spin label to estimate the size of EcFeoB at ≈300 kDa. The results suggest that EcFeoB can form higher-order oligomers under these conditions (Hagelueken et al. 2015). However, spin labelled, full-length EcFeoB solubilized in DDM or incorporated into 3-([3-Cholamidopropyl]dimethylammonio)-2-hydroxy-1-propanesulfonate/dipalmitoylphosphatidylcholine (CHAPSO/DMPC) bicelles did not appear to form higher order oligomers based on PELDOR experiments. These results were confirmed by mass spectrometry and negative-stain electron microscopy (Hagelueken et al. 2016). Thus, the in vitro oligomerization of EcFeoB remains an open debate.

In 2018, expression, solubilization, and purification of full-length Klebsiella pneumoniae FeoB in DDM and C₁₂E₈ was the first in vitro study of this construct. In contrast to previous studies that measured the GTP hydrolysis rates of KpNFeoB alone, GTP hydrolysis of full-length KpFeoB was determined to be orders of magnitude higher at ≈0.09 s⁻¹ for the DDM-solubilized protein and ≈0.03 s⁻¹ for the C₁₂E₈-solubilized protein (Smith and Sestok 2018). The stimulation of GTP hydrolysis by exogenous potassium was not observed, in contrast to KpNFeoB. Intriguingly, given the rapid rate of basal GTP hydrolysis, these observations suggest that the TM region of FeoB is able to significantly increase GTP hydrolysis levels, which could be operative for active transport. Whether FeoB transports Fe²⁺ in a facilitated or active manner remains an open debate. Additional factors that could stimulate GTP hydrolysis further, such as protein-membrane or protein-protein interactions, have yet to be identified (Smith and Sestok 2018). This paucity of information on the intact, full-length protein further emphasizes the need for additional biophysical and biochemical studies of intact FeoB.

3.5.4 FeoA

Within the feo operon, feoA is the most common gene appearing alongside feoB (Lau et al. 2013), suggesting FeoA plays an important role in Fe²⁺ uptake. FeoA is a small, cytoplasmic, β-barrel protein (Fig. 12.7) of approximately 8 kDa chiefly composed of a Src Homology 3-Like (SH3) domain. SH3 proteins fold to form a hydrophobic core, with the β-sandwich held together by three loops termed the RT-Src loop, the N-Src loop, and the distal loop. The RT-Src loop and the N-Src loop contribute to peptide binding while the distal loop contributes to ligand binding (D’Aquino and Ringe 2003; Li 2005). In eukaryotes, SH3 domains have been implicated in a suite of signal transduction mechanisms, facilitating protein-protein interactions by interacting with their binding partners through a Pro-rich region that typically folds into a left-handed helical conformation (D’Aquino and Ringe 2003; Li 2005). SH3 proteins were initially believed to be absent from prokaryotes, but have since been discovered while remaining widely uncharacterized (D’Aquino and Ringe 2003). Structures of FeoA from different organisms have been determined, yet the function of FeoA still remains unknown. Two plausible roles for FeoA include functioning as either a GEF or as a GAP (Fig. 12.8). In eukaryotes, GEFs and GAPs are crucial regulatory proteins that participate in the nucleotide cycling of G-proteins and provide signaling specificity (Bos et al. 2007; Vetter and Wittinghofer 2001). GEFs function by physically removing GDP bound to G-proteins via protein-protein interactions, allowing for GTP subsequently to bind to the apo G-protein (turning the signal from “off” to “on”). Conversely, GAPs function by catalyzing the hydrolysis of GTP to GDP, most commonly by intercalating a positively-charged residue on the GAP (such as Arg and Lys) into the GTP-binding site on the G-protein, increasing the polarization rate of the GTP phosphoester bond (turning the signal from “on” to “off”) (Fig. 12.8) (Bos et al. 2007; Vetter and Wittinghofer 2001). Whether FeoA functions as a GEF or GAP remains to be seen, but the structure of FeoA strongly suggests its involvement in protein-protein interactions.

Fig. 12.7

X-ray crystal structure of KpFeoA (PDB ID: 6E55). (a) FeoA is a β-barrel protein composed of an SH3-like domain. (b) FeoA contains a hydrophobic clamp with Leu resides (displayed in gray) postulated to be involved in protein-protein interactions with the N-terminal domain of FeoB. Panel b is displayed as a 90° vertical rotation relative to Panel a. ‘N’ and ‘C’ represent the N- and C-termini respectively. Figure created with BioRender

Fig. 12.8

Cartoon representing two possible functions of FeoA (red polygon) with respect to NFeoB (teal polygon). FeoA has been speculated to act as either a GTPase activating protein (GAP), which facilitates GTP hydrolysis to GDP in order to turn “off” function, or as a guanine exchange factor (GEF), which facilitates exchange of bound GDP for GTP in order to turn “on” function

While several unpublished NMR structures of the FeoA SH3-like domain appeared in the PDB (Protein Data Bank) in the early 2000s (P. aeruginosa FeoA PDB ID: 2H3J; K. pneumoniae FeoA PDB ID: 2GCX; and Clostridium thermocellum FeoA PDB ID: 2K5L) the first published X-ray structure of FeoA appeared in 2010 from Stenotrophomonas maltophilia (PDB ID: 3MXH) (Su et al. 2010). The structure was determined to 1.7 Å resolution and contained two molecules in the asymmetric unit. Both molecules of FeoA adopted a SH3-like fold, as seen in the previous, unpublished NMR structures (Su et al. 2010). The SH3-like fold in FeoA is chiefly composed of 5 β-strands linked together through the RT-loop, the N-Src-loop, and α-helices (Fig. 12.7). One intriguing aspect of the SmFeoA structure is the presence of two zinc ions and six chloride ions, which the authors attribute to the facilitation of SmFeoA oligomerization (Su et al. 2010). The second published structure of FeoA was an NMR structure of EcFeoA (Lau et al. 2013). Similar to SmFeoA, EcFeoA comprises two antiparallel β-sheets that form the β-barrel, and unstructured loop regions. However, SmFeoA contains three α-helices whereas EcFeoA only contains 2 α-helices (Lau et al. 2013; Su et al. 2010). These small, structural differences could be the result of low sequence conservation between SmFeoA and EcFeoA, while EcFeoA and KpFeoA have high sequence identity (90%) and are more structurally similar (Lau et al. 2013; Su et al. 2010). Finally, a recent X-ray crystal structure of KpFeoA (PDB ID: 6E55) was determined to 1.5 Å resolution. Noteworthy was the observation of FeoA-FeoA interactions, which were speculated to be the location for FeoA-NFeoB interactions. Two sets of KpFeoA dimers were present in the asymmetric unit and participate in unique FeoA-FeoA interactions at the dimer interface via intercalation of hydrophobic Leu residues (Fig. 12.7b) from one KpFeoA molecule with Ala residues on the second KpFeoA molecule (Linkous et al. 2019). When compared to an unpublished NMR structure of KpFeoA (PDB ID: 2GCX), there is an ≈4 Å closure in the clamp. This crystallographic KpFeoA is believed to represent the “closed” state while the FeoA present in the NMR structure is believed to represent the “open” state. In silico docking experiments using a 10 amino acid sequence from KpFeoB containing the PXXP motif revealed that the 10-mer could bind in the “open” KpFeoA model. Strikingly, the 10-mer was predicted to bind within the “C-shaped” clamp (Fig. 12.7b), similar to where the FeoA-FeoA interactions occur, but only in the “open” form of the model. It is thus possible that FeoA interacts with FeoB at the PXXP site via the hydrophobic residues in the clamp region and could alter the nucleotide status of the G-protein domain by increasing GTP hydrolysis or facilitating GDP release thus regulating Fe²⁺ uptake via FeoB (Linkous et al. 2019). However, these structures alone are insufficient to determine the function of FeoA and more research will be necessary to determine the nature of the interaction between FeoA and FeoB and whether these hydrophobic residues are necessary to mediate the interaction.

Despite the availability of several FeoA structures revealing an SH3-like fold, the function of FeoA remains unknown and is of debate. Unlike eukaryotic SH3 domains, prokaryotic SH3 domains are not as well characterized. The presence of the SH3-like domain suggests that FeoA could be involved in protein-protein interactions, likely with FeoB as suggested by the KpFeoA structure. However, while some studies have tried to establish the function of FeoA little information exists about FeoA-FeoB interactions at the protein level. One study explored the effect of EcFeoA on the GTP hydrolysis activity of EcNFeoB by using NMR to monitor ³¹P signals indicative of GTP hydrolysis (Lau et al. 2013). In the presence of equimolar amounts of EcFeoA, EcNFeoB did not exhibit any obvious changes in GTPase activity, leading to the conclusion that FeoA does not act as a GAP on NFeoB alone (Lau et al. 2013). However, the effect of FeoA on intact FeoB is still unknown and higher stoichiometric ratios of FeoA:FeoB may be necessary for activity. More studies will ultimately be necessary to determine if FeoA has any effect on GTP hydrolysis, GDP release, and/or Fe²⁺ transport.

3.5.5 FeoAB Fusions

Some feo operons encode for a single polypeptide in which FeoA is naturally fused to the N-terminal soluble domain of FeoB. While these fusions are rare in genomic distribution, these observations further support the hypothesis that FeoA and FeoB are meant to interact physically to affect function. While these fusions are not well studied, they appear in the genomes of some very common human pathogens, such as Porphyromonas gingivalis and Bacteroides fragilis. Work on these fusions has chiefly been done at the cellular level (vide infra); unfortunately, no work at the protein level has examined these proteins.

3.5.6 FeoC

FeoC is a small protein of unknown function encoded along the feoC portion of the feo operon (Fig. 12.9). Unlike the strongly conserved feoA gene, feoC is thought to be present in only ≈13% of feo operons, predominantly in γ-proteobacteria (Lau et al. 2013). The solution NMR structure of monomeric KpFeoC (PDB ID: 2K02) was the first to be described and demonstrated that FeoC contains an N-terminal helical domain and a C-terminal unstructured loop, which combined are reminiscent of a winged-helix domain (Fig. 12.9) (Hung et al. 2012a). It has therefore been suggested that FeoC could serve as a transcriptional regulator (Cartron et al. 2006). However, upon comparison of the protein electrostatics of KpFeoC to that of other DNA binding proteins, such as DtxR and the DNA binding domain of BlaI, it has been noted that the corresponding helix in KpFeoC is very negatively charged and therefore would not favor an interaction with DNA based on electrostatics. Supporting this statement, the authors failed to detect DNA binding to KpFeoC (Hung et al. 2012a). Furthermore, the authors also observed Zn²⁺ binding (but surprisingly not Fe²⁺ binding) in the winged region of KpFeoC, although this failed to promote DNA binding (Hung et al. 2012a). While KpFeoC did not bind Fe²⁺, the presence of four conserved Cys residues in the wing region (including a CXXC motif) suggested the possibility of iron-sulfur cluster binding (Fig. 12.9). As such, cellular and biochemical studies have sought to understand the identity and the role of iron-sulfur cluster binding to FeoC.

Fig. 12.9

Lowest-energy NMR conformer of E. coli FeoC (PDB ID: 1XN7). Displayed in gray and yellow are the highly-conserved Cys residue sidechains that are involved in coordination of the [Fe-S] cluster. Labels ‘N’ and ‘C’ represent the N- and C-termini, respectively

The determination of the exact nature of the FeoC [Fe-S] cluster was difficult due to its redox- and oxygen-sensitivity. The first biochemical study to investigate the ability of KpFeoC to bind an [Fe-S] cluster was published in 2013. The presence of a paramagnetic center in KpFeoC was first indicated by strong paramagnetic shifts in ¹H NMR experiments. Further biophysical characterization of reconstituted KpFeoC (exhibiting only 10% incorporation) under oxic conditions using electronic absorption spectroscopy suggested that the identity of the cluster was either that of a [2Fe-2S] cluster or a [4Fe-4S] cluster (Hsueh et al. 2013). When reconstituted KpFeoC was analyzed by electron paramagnetic resonance (EPR) spectroscopy, experimental g values were attributed to the presence of an unusual high-potential iron sulfur protein (HiPIP) cluster (Hsueh et al. 2013). Lastly, X-ray absorption spectroscopy (XAS) confirmed the cluster in KpFeoC to be ligated by four cysteine residues (Hsueh et al. 2013). The role of the cysteine residues in KpFeoC were evaluated by testing the ability of cysteine to serine variants to bind an iron-sulfur cluster. Three of four of the cysteine residues (Cys⁵⁶, Cys⁶⁴, and Cys⁷¹) were essential for cluster binding whereas Cys⁶¹ was not essential (Hsueh et al. 2013). Very surprisingly, the [4Fe-4S] cluster present in KpFeoC had an unusually long average half-life of 17 h upon exposure to oxygen (Hsueh et al. 2013).

The most recent publication that characterized the iron-sulfur cluster in the FeoC protein was published in 2019. In this study, the cluster was reconstituted under strictly anoxic conditions and subsequently characterized by electronic absorption spectroscopy, EPR, XAS, and dynamic light scattering (DLS) (Smith et al. 2019). Reconstituted under anoxic conditions, the cluster of EcFeoC was assigned to the more common redox active [4Fe-4S]^2+/+ species (Smith et al. 2019). Electronic absorption spectra of the cluster-bound EcFeoC revealed that the [4Fe-4S]²⁺ cluster is O₂-sensitive and rapidly degrades to a [2Fe-2S]²⁺ cluster in ≈5 min (k_obs ≈ 0.04 s⁻¹) in O₂-replete conditions (Smith et al. 2019). When these experiments were repeated for KpFeoC, nearly identical electronic absorption features, XAS spectra, and EPR signals were observed confirming that KpFeoC also binds a redox active, O₂-sensitive [4Fe-4S]^2+/+ cluster (Smith et al. 2019). Lastly, anoxic DLS measurements demonstrated that cluster binding to both EcFeoC and KpFeoC resulted in a compaction of the protein conformation, but cluster binding did not promote oligomerization. Though the exact role of FeoC remains unknown, these data support that FeoC may function as a redox-active iron sensor, similar to that of FNR (Smith et al. 2019). However, some bacteria, such as V. cholerae lack the necessary Cys residues that are required to bind an [Fe-S] cluster. Whether these proteins function similarly to cluster-binding FeoCs has yet to be determined.

3.5.7 FeoA/B/C Interaction Studies

Despite in vitro studies of small soluble proteins of the Feo system (FeoA and FeoC) and single isolated domains (NFeoB), few studies have investigated how FeoA, FeoB, and/or FeoC may function in concert to transport Fe²⁺ iron across a bacterial membrane. In 2012, the ability of FeoA and FeoB to interact was explored in Salmonella enterica using a bacterial two hybrid (BACTH) assay. When feoB was deleted from strain JH380 (ΔmntH, ΔsitABCD background), Fe²⁺ uptake was threefold lower than in the WT, and these results were recapitulated with an feoA deletion. Reintroduction of feoA was able to rescue the phenotype. In other words, the loss of feoA was as detrimental to ferrous iron uptake in S. enterica as the loss of feoB (Kim et al. 2012). An feoAB deletion strain was then created to test whether FeoA function would be dependent on FeoB. Complementation with only feoB resulted in Fe²⁺ uptake approximately three-fold lower than JH380, whereas complementation with only feoA resulted in no rescue of Fe²⁺ uptake; complementation of both feoA and feoB restored Fe²⁺ uptake. These results suggest that FeoA is necessary for FeoB to transport ferrous iron in vivo (Kim et al. 2012).

As FeoA-based dependence might be the result of protein-protein interactions between FeoA and FeoB, this hypothesis was further probed in vivo. A BACTH assay in S. enterica where FeoA was fused to T18 and FeoB was fused to T25 was initially constructed, and β-galactosidase activity indicated an interaction between FeoA and FeoB in vivo (Kim et al. 2012). FeoA variants (P20A/L26Q and W9G/L26Q) demonstrated ≈28-fold lower β-galactosidase activity compared to the WT. Single variant studies suggested that L26 may be an essential residue for FeoA-FeoB interactions (Kim et al. 2012). These cellular results support the hypothesis that FeoA is important for FeoB-mediated ferrous iron transport in S. enterica (Kim et al. 2012), but the exact mechanism is unknown.

The presence of the FeoC HTH motif suggests that the protein could serve as a transcriptional regulator of the Feo system, which could be linked to the protein’s cluster composition. As such, studies have examined how FeoC affects the expression levels of the feoB gene in S. enterica (Kim et al. 2013). Maximal expression of feoB occurs under both low iron and low O₂ conditions, which could be further enhanced in the absence of a functional fur gene or impaired in the absence of a functional fnr gene (Kim et al. 2013). A feoC deletion strain exhibited low levels of feoB expression, which could be restored by complementation with an feoC-containing plasmid. Interestingly, levels of feoB mRNA were higher in this strain than in the WT suggesting that feoC does not act at the point of transcription, but rather post-transcriptionally (Kim et al. 2013). BACTH experiments demonstrated that FeoB and FeoC interact, and these results were confirmed using pulldown experiments with NFeoB and His₆-FeoC, although the oxidation state and metal-bound form of FeoC was not controlled. S. enterica harboring both a feoC and a ftsH deletion resulted in lower levels of FeoB suggesting that FeoC may protect FeoB from proteolysis by FtsH. Additionally, a ΔfeoC strain of S. enterica accumulated 4.5-fold less Fe²⁺ than the WT strain indicating that the FeoC protein is important for Fe²⁺ uptake in vivo (Kim et al. 2013). A follow-up study investigated Lon-mediated proteolysis of FeoC in S. enterica (Kim et al. 2015). FeoC expression from an IPTG-inducible plasmid in S. enterica was only detectable under low-oxygen conditions and did not appear to be affected by low- or high-iron concentrations. Interestingly, the FeoC protein was rapidly degraded when expressed in cells growing in a high-oxygen and high-iron environment (t_½ ca. 5 min), similar to the rate of the O₂-mediated cluster decomposition. A switch to anoxic conditions increased FeoC stability (t_½ ca. 50 min) and was unaffected by the presence of absence of iron (Kim et al. 2015). Mutation of all four Cys residues in the wing of FeoC, the likely sites for binding of an iron-sulfur cluster as shown in K. pneumoniae and E. coli, further increased FeoC stability. Additionally, FeoC stability appears to affect the accumulation of FeoB. In the absence of the Lon-protease, FeoC can accumulate under both high- and low-oxygen conditions which promotes the accumulation of FeoB (Kim et al. 2015). The authors suggest that this could be a method by which S. enterica regulates Feo-mediated Fe²⁺ uptake under anoxic and iron-limiting conditions, when regulation but FUR and FNR would be negligible (Kim et al. 2015). Binding of an iron-sulfur cluster to FeoC could serve to control the oxygen-sensitivity of FeoC; however, no studies have examined how cluster binding affects FeoC stability and degradation.

In Vibrio cholera, the causative agent of the disease cholera, the Feo system is composed of FeoA, FeoB, and FeoC, which may interact to regulate Feo-mediated Fe²⁺ transport (Weaver et al. 2013). A BACTH assay was performed to assess interactions between and among the Feo proteins. No interactions were detected between FeoA-FeoB or FeoA-FeoC. In contrast, this method did reveal an interaction between FeoB and FeoC, which was mapped to the N-terminal domain of FeoB. E29G and M35A variants of FeoC abolished the FeoB-FeoC interaction, but conservation of these residues is poor among γ-proteobacteria (Weaver et al. 2013). In contrast to these first studies, later work on the V. cholerae Feo system suggested that FeoA, FeoB, and FeoC all interact to form a large complex. In 2016, in vivo formaldehyde cross-linking was used to probe which Feo complexes might be present or active in vivo. Several higher-order complexes of FeoA, FeoB, and FeoC were observed localized to the inner membrane based on SDS- and blue native-PAGE. The complexes were estimated to be ≈250 kDa, ≈500 kDa, and ≈720 kDa. Immunoprecipitation and LC-MS/MS demonstrated that FeoA and FeoB could be detected in the complex at ≈720 kDa, but not FeoC. However, when the V5 tag for immunoprecipitation was placed on FeoC instead of FeoB, then FeoC could be detected in the largest complex suggesting that the tag may hinder FeoB-FeoC interaction (Stevenson et al. 2016). The exact stoichiometry of the large complex (≈720 kDa) could not be determined, but it was suggested that a trimer of FeoB trimers may interact with one or more FeoA or FeoC proteins to transport ferrous iron (Stevenson et al. 2016). Some additional complexes were found to contain only FeoB, while a complex at ≈100 kDa was found to contain FeoA, FeoB, and FeoC. Smaller complexes were proposed to be intermediates or possibly products from the disassembly and the breakdown of a larger Feo complex. Additionally, when FeoA was tagged with the V5 epitope for immunoprecipitation, the protein could not be detected in any complexes (Stevenson et al. 2016). It could not be determined which, if any, of these complexes are dominant within the cell, but these results do suggest that all three of the Feo proteins may interact at some point within the cellular context.

Mutational analyses have given some insight into how a larger Feo complex may form in V. cholerae. For example, an FeoB K15D variant in the G1 motif abolished Feo function and resulted in no detectable complex formation, indicating that nucleotide binding may be important for protein-protein interactions. An FeoB variant of the switch II region, D72A, resulted in the loss of function and the inability to form a larger complex (Stevenson et al. 2016). In an FeoC deletion strain, FeoB was found expressed at lower levels in the inner membrane, yet an FeoA-FeoB complex was still observed, suggesting that FeoC plays a role in FeoB expression but is not required for complex formation. In contrast, plasmid-driven expression of FeoB in tandem with any of four FeoA variants (G32K, A45D, P50R, and V72K) failed to display complex formation, suggesting that intact FeoA is a pre-requisite for Feo interactions (Stevenson et al. 2016). Further structural studies on Feo protein complexes would be useful in clarifying why and how these variants lead to disruption in complex formation.

Finally, structural and biophysical measurements have shown an interaction between Klebsiella pneumoniae NFeoB and its cognate FeoC. ITC experiments of nucleotide-free KpNFeoB and apo KpFeoC titrations demonstrated a 1:1 molar interaction with a K_d of ≈0.5 μM, and SEC experiments revealed co-migration of monomeric, nucleotide-free KpNFeoB with apo, monomeric KpFeoC (Hung et al. 2012b). These observations were supported by an X-ray crystal structure of nucleotide-free KpNFeoB co-crystallized with apo KpFeoC. Each asymmetric unit contained one nucleotide-free KpNFeoB and one apo KpFeoC in which the N-terminus of FeoC interacted with the GDI domain of NFeoB (Hung et al. 2012b). Unfortunately, the wing region of FeoC was not present in electron density, likely a result of the flexibility of the wing region (Hung et al. 2012b). Additionally, this dynamic region of KpFeoC is known to interact with an [Fe-S] cluster, which was not probed. Thus, it is unclear whether either [Fe-S] cluster binding or possibly nucleotide binding would affect this NFeoB-FeoC interaction and how.

4 Metal-Rich, Anoxic, and Acidic Environments

Iron-oxidizing bacteria (FeOB) utilize the oxidation of Fe²⁺ to Fe³⁺ as a means to generate energy for growth. These organisms obtain the iron necessary for essential metabolic processes from their native environment and play an important role in the iron biogeochemical cycle (Emerson et al. 2010). Given the insolubility of Fe³⁺, and the production of ROS upon spurious Fe²⁺ oxidation, iron-dependent survival presents a unique challenge to FeOB. Furthermore, the energy derived from abiotic Fe²⁺ oxidation is low (≈29 kJ mol⁻¹) near neutral pH, and the half-life of Fe²⁺ under oxic conditions is <1 min (Emerson et al. 2010; Roden et al. 2004). FeOB living under these conditions would therefore produce iron oxyhydroxides, which would be detrimental to the organism. As a result, FeOB must grow under microaerophilic conditions that stabilize Fe²⁺ over Fe³⁺ (Emerson et al. 2010; Roden et al. 2004). FeOB inhabit a variety of ecological niches including the soil, freshwater and marine environments, waste and bioreactor sites, and hydrothermal vents (Emerson et al. 2010). A greater understanding how FeOB transport Fe²⁺ would thus improve our understanding of their contribution to the iron biogeochemical cycle and would shed light on this ancient form of metabolism.

In contrast to FeOB, iron-reducing bacteria (FeRB or IRB) generate energy by reducing Fe³⁺, typically from insoluble iron oxides, to Fe²⁺. Bacteria that couple this process to the oxidation of organic molecules, such as sugars and amino acids, aromatic compounds, long chain fatty acids, and butyrate, propionate, and acetate or H₂ are termed dissimilatory iron reducing bacteria (Esther et al. 2015; Lovley 1997; Richter et al. 2012; Weber et al. 2006). These organisms can often be found cohabitating with FeOB because the Fe³⁺ produced may function as an FeRB feedstock. FeRB play a role in mineral formation, such as when reduced iron is exposed to oxygen, and iron cycling (Lovley 1997; Esther et al. 2015; Richter et al. 2012; Weber et al. 2006). Reduction of Fe³⁺-rich surfaces can be accomplished via several routes. One of these routes is by direct contact of a bacterial biofilm with an Fe³⁺-rich surface, as the biofilm may help mediate electron transfer. Fe³⁺ reduction can also be facilitated by chelators or siderophores that solubilize Fe³⁺ and increase its bioavailability (Esther et al. 2015; Lovley 1997; Thormann et al. 2004; Richter et al. 2012; Weber et al. 2006). Additionally, FeRB can cycle electron shuttles to reduce Fe³⁺-rich surfaces. Somewhat similar to biofilm contact, pili, or protein nanowires, can also mediate contact with Fe³⁺-rich surfaces for reduction (Esther et al. 2015; Richter et al. 2012; Brutinel and Gralnick 2012; Weber et al. 2006). Unfortunately, the iron transport mechanisms in these bacteria remain largely uncharacterized, but a synopsis of what is currently known in the literature is presented below.

4.1 Magnetotactic Bacteria

Magnetotactic bacteria (MB), microorganisms that orient themselves and move along Earth’s magnetic field through a process known as magnetotaxis, were first discovered in 1975 (Blakemore 1975). MB were isolated from samples taken from surface sediments of salt marshes and surface layers of sedimentary cores in Massachusetts (Blakemore 1975). Using transmission electron microscopy (TEM), crystal-like particles that are now termed magnetosomes were initially discovered. At that time, it was posited that these intracellular compartments might be magnetic, formed from a mineral such as magnetite. Supporting this presumption, energy dispersive X-ray microanalysis subsequently revealed these particles to be composed predominantly of iron. Concurrently, different species of morphologically distinct MB were also identified in several other marsh muds (Blakemore 1975). In 1988, MB isolated from an estuarine salt marsh were analyzed for elemental content of the magnetic particles, which were determined to be composed of iron and oxygen in the form of magnetite (Fe₃O₄) constituting ca. 1.6% of the dry weight of the organism (Bazylinski et al. 1988). Variations in particle composition, comprising iron and sulfur in the form of greigite (Fe₃S₄) and pyrite (FeS₂), were then found (Mann et al. 1990) and can be correlated to distinct growth environments.

MB are diverse microorganisms that grow optimally within anoxic and microaerophilic environments, but they can also be isolated from deep sea sediments. In the deep oceans, MB are proposed to play an important role in marine iron and sulfur cycling as accumulated, intracellular iron and sulfur can be released into the environment upon cell death and lysis thus promoting additional microbial activity (Simmons et al. 2004). High concentrations of particulate iron support a high abundance of MB, and this environment occurs at the oxycline, the point in the aquatic environment where the concentration of oxygen changes steeply and beyond which sulfur concentrations begin to peak (Simmons et al. 2004). MB appear to be widely distributed among the bacterial kingdom, including the classes α- through ε-proteobacteria, Chlorobiales, and Cyanobacteria, among others (Simmons et al. 2004). Due to their high iron requirements and their anoxic and microaerophilic growth environments, it is likely that the iron acquisition systems present in MB mimic those present on early Earth, providing a glimpse of life prior to the GOE.

Consistent with their high demand for iron, MB must utilize a combination of tactics to accumulate this essential element. One major strategy is the use of siderophores. Like MB, many (but not all) prokaryotes engage in the metabolically-demanding biosynthesis and secretion of siderophores to scavenge for ferric iron. Retrieved ferric iron trapped in these siderophores can be released by either destruction of the siderophore itself via bond cleavage, or through a ferric reductase-mediated iron reduction and release mechanism (Cain and Smith 2021). Importantly, as MB are found in a number of anoxic niches, ferrous iron acquisition also represents a major acquisition route of reduced environmental iron.

Magnetospirillum magneticum strain AMB-1 relies on catechol and hydroxamate siderophores as well as ferrous iron transport as its primary means of iron uptake (Calugay et al. 2003). A global expression analysis of the M. magneticum strain AMB-1 genome has revealed multiple iron-regulated genes in this organism. In magnetosome-forming cultures, approximately 70% of extracellular iron was rapidly assimilated within 1 h (Suzuki et al. 2006). Under these conditions Fe²⁺ transport genes (ftr1, tpd, feoA, and feoB) were upregulated whereas in non-magnetosome-forming conditions ferric iron uptake genes (such as tonB, fepA, and napABC) were down-regulated (Suzuki et al. 2006). In contrast to the down-regulated ferric iron uptake genes, cirA and fepC were upregulated in iron-replete conditions and encode a ferric-siderophore outer membrane receptor and an inner membrane ferric-siderophore transporter, respectively (Suzuki et al. 2006). M. magneticum strain AMB-1 is capable of growing under oxic conditions (Matsunaga et al. 1991), and at least one catechol siderophore has been identified to be 3,4-dihydroxybenzoic acid (Calugay et al. 2006). Siderophore-mediated ferric iron uptake was significant when concentrations of ferric iron in the medium were 40–80 μM (Calugay et al. 2003), but this condition is not conducive to magnetosome formation.

When taken together, these results indicate that Fe²⁺ uptake is likely the predominant source of iron under magnetosome-forming conditions (Suzuki et al. 2006), and there seems to be more than one route for ferrous iron assimilation. Ftr1 is uncharacterized but annotated as a Fe²⁺/Pb²⁺ permease, Tpd is uncharacterized but annotated to be involved in high-affinity Fe²⁺ uptake, and Amb3335 is predicted to be a ferric iron reductase, which is also upregulated in iron-replete conditions (Suzuki et al. 2006). In a nonmagnetic mutant of M. magneticum strain AMB-1 deficient for siderophore uptake (denoted strain NMA61), a cytoplasmic ATPase that contributes to Fe²⁺ uptake was also identified (Suzuki et al. 2007). This ATPase bears homology to ArgK, a protein kinase (Suzuki et al. 2007). Previous reports speculate that the ATPase may provide energy for a Fe²⁺ transporter such as Feo, Tpd, and Frt1, thus supplying ferrous iron to magnetosome formation (Suzuki et al. 2007). In 2008, iron transporters expressed in M. magneticum strain MS-1 were identified using 2-D electrophoresis by comparing proteins expressed during growth in Fe²⁺- and Fe³⁺-rich media (Taoka et al. 2009). Two of the proteins expressed under Fe²⁺-rich conditions were 76 kDa and 70 kDa in size and with homology to TonB-dependent outer membrane ferric-siderophore receptors in Beijerinckia indica and Rhodopseudomonas palustris, respectively (Taoka et al. 2009). Two sets of the feoA and feoB genes were also identified in this study. Expression of FeoB was confirmed by utilizing polyclonal antibodies generated from recombinant NFeoB1. In these studies, FeoB appears as a band at ≈75 kDa localized to the cytoplasmic membrane, consistent with predictions of FeoB’s location (Taoka et al. 2009). Further functional studies have yet to be reported.

M. gryphiswaldense MSR-1 does not secrete siderophores but does utilize Fe³⁺ through extensive ferric reductase activity and subsequent Fe²⁺ uptake. Ferric reductase activity in whole cells, the cytoplasm, and membrane fractions (but not the periplasm) of M. gryphiswaldense was first demonstrated in 2007 (Xia et al. 2007). Native PAGE coupled with ferric reductase active staining afforded the identification of six bands in the cytoplasmic fraction and three bands in the membrane fraction of cells corresponding to ferric reductase isozymes. These ferric reductases were named FeR-1 to FeR-6. FeR-6 was subsequently purified as it had the greatest activity among all of the reductases. FeR-6 is ≈16 kDa and does not appear to have sequence homology to other proteins (Xia et al. 2007). Ferric citrate, reduced nicotinamide adenine dinucleotide (NADH), and oxidized flavin mononucleotide (FMN) were used in ferric reductase assays to determine the reductase activity of FeR-6. The protein had a modest K_M for ferric citrate of ≈45 μM and a V_max of ≈1.2 μM/min. Interestingly, ferric reductase activity was strongly inhibited by Ag⁺ and divalent metal ions such as Zn²⁺, Mn²⁺, Cu²⁺, and Co²⁺ (Xia et al. 2007). In a follow-up study, FeR-5 and FeR-6 were identified as bifunctional enzymes with thioredoxin reductase activity and flavin reductase activity, respectively, and deletion of both genes inhibited magnetosome formation (Zhang et al. 2013). Although FeR-6 could reduce Fe³⁺ without the FMN cofactor, Fe³⁺ reductase activity was 20-fold higher with FMN (Zhang et al. 2013), emphasizing the importance of the flavin cofactor. The remaining four isozymes have not yet been characterized.

As the ferric reductases are located in the cell membrane, they are speculated to provide reduced iron for transport via a membrane Fe²⁺ transporter (Zhang et al. 2013). Early studies examining the growth and magnetosome formation of M. gryphiswaldense MSR-1 indicate that magnetite formation can only occur under microaerophilic conditions and is tightly coupled to iron uptake (Schüler and Baeuerlein 1998). Under these conditions, recent experiments suggest that ferric reductase activity could supply the inner membrane transporter FeoB with Fe²⁺, which is then transported into the cell for magnetosome incorporation. In 2008, two feoB genes in M. gryphiswaldense strain MSR-1 were identified and characterized (Rong et al. 2008). The predicted FeoB1 protein encoded by the feoB1 gene (704 amino acids) has high sequence identity to FeoBs from other magnetotactic bacteria but only 35% identity (53% similarity) to FeoB2, encoded by feoB2. A putative feoA gene was found within the region of feoB1 and named feoA1. FeoB1 contains conserved G-protein motifs in the N-terminal region, and the C-terminal domain was predicted to be imbedded in the cytoplasmic membrane by 9 TM helices (Rong et al. 2008). TEM was used to monitor magnetosome formation of a ΔfeoB1 strain, which produced lower amounts of magnetosomes, smaller magnetosomes, and lower iron content compared to the WT strain. These defects could be rescued by complementation with feoB1. Similar results were observed for cells grown in the presence of either Fe²⁺- or Fe³⁺-citrate (Rong et al. 2008). The feoA1 and feoB1 genes were also found to be downregulated under iron-rich conditions. A later study investigated the role of feoB2 in M. gryphiswaldense MSR-1 (Rong et al. 2012). Interestingly, when a ΔfeoB2 strain was analyzed for magnetosome formation, the size and number of the magnetosomes were similar to the WT strain suggesting that FeoB2 does not contribute to magnetosome formation when FeoB1 is present. Surprisingly, in cells lacking both a functional FeoB1 and FeoB2, magnetosome formation and size was similar to that of WT; however, the number of magnetosomes per cell were lower in the double mutant than in the feoB1 deletion strain. Additionally, when cells were grown in the presence of ferric citrate, the ΔfeoB2 strain had significantly lower levels of cellular iron content than that of the ΔfeoB1 strain (Rong et al. 2012). For all deletion strains, the cellular metal content of Mn, Zn, Cu, and Mg was similar to WT confirming that both FeoBs are dedicated to iron uptake (Rong et al. 2012). The deletion strains were also tested for their ability to respond to oxidative stress. Deletion strains demonstrated sensitivity when grown in 500 μM H₂O₂, and all had significantly lower superoxide dismutase and catalase activities when compared to the WT strain (Rong et al. 2012), suggesting that some of the acquired ferrous iron is necessary for assimilation into these two essential enzymes that combat oxidative stress.

4.2 Gallionella and Geobacter

Gallionella ferruginea, a microaerophilic bacterium with a growth requirement for Fe²⁺, was first described in 1837 and is the first known FeOB to be identified. G. ferruginea has an intriguing ability to form an Fe-oxyhydroxide-encrusted stalk during growth (Emerson et al. 2010). G. ferruginea is able to grow at the end of the stalk, which can form at a rate of 80–90 μm/h in microcultures. These stalks are likely used as a positioning mechanism to find the appropriate iron/oxygen gradients needed for G. ferruginea growth, but they may also function as a sink to deposit precipitated iron oxyhydroxides that would otherwise result in the death of the organism (Hanert 1974; Emerson et al. 2010). Questions remain regarding the composition of the stalks and the mechanism by which the precipitated iron is excreted.

While Fe²⁺ transport mechanisms have not been well-characterized in Gallionella, this transport process undoubtedly contributes to the ability of this organism to cycle iron in different and often harsh iron-rich environments. Acid mine drainage (AMD) sites are low-pH, anoxic springs containing high levels of iron as a result of mining (Jones et al. 2015). The Upper and Lower Red Eyes in Pennsylvania constitute an AMD with pH ranging from 4.0–4.5 and Fe²⁺ concentrations >6.5 mM. FeOB, such as Gallionella, offer the possibility to bioremediate such sites using biological iron oxidation (Jones et al. 2015). An analysis of free-living bacteria at this site showed that the Gallionellaceae family composed 42% of all bacterial species identified. Bacteria from the orders of Rhodospirillales, Acidimicrobiales, Xanthomonadales, Acidobacteriales, and Nitrospira were also present, but their distribution was dependent on both pH and [Fe²⁺]. For instance, Gallionellaceae were more abundant at high Fe²⁺ concentrations while Acidithiobacillus spp. were more abundant at lower Fe²⁺ concentrations (Jones et al. 2015). Though not as well characterized, the Arctic tundra also contains regions where FeOB inhabit acidic soils, sediment surfaces, and water sources. Gallionella spp. have been identified in all of these environments (Emerson et al. 2015). Thus, it is clear that adaptation to high acidic environments that may be rich in Fe²⁺ is necessary to support the survival of Gallionella and related species.

Similarly adapted is Geobacter sulfurreducens, a hydrogen- and acetate-oxidizing, dissimilatory metal- and sulfur-reducing bacterium that can be found in soils, aquatic sediments, and subsurface environments. Reduction of insoluble minerals such as Fe³⁺ oxides is facilitated by electron transport through pili rich in c-type cytochromes, which are localized to the periplasm and often excreted into the extracellular space (Caccavo et al. 1994; Lovley and Walker 2019; Weber et al. 2006; Smith et al. 2013; Seeliger et al. 1998). Though iron transport has not been extensively studied in Geobacter, one study analyzed the Geobacter iron stimulon by determining differences in gene expression for cells growing in iron-replete, iron-sufficient, and iron-deficient conditions. Genes differentially expressed in Geobacter encoded for c-type cytochromes and other proteins containing heme or iron-sulfur clusters, and Fe³⁺ reductases such as the OmcZ cytochrome (Embree et al. 2014). Both FeoA and FeoB were the most downregulated during Fe³⁺ reduction, suggesting that Fe²⁺ generated from reduction is not used as a substrate for the Feo system. Eleven efflux pumps/subunits and one ferritin-like protein domain were also downregulated, while another ferritin-like protein domain was upregulated (Embree et al. 2014). The Fur and IdeR proteins were more highly expressed under iron-sufficient conditions and exhibited repression as extracellular Fe²⁺ concentrations increased. These results suggest that iron homeostasis is tightly regulated in Geobacter in response to the large amounts of Fe²⁺ produced from metabolic processes and to maintain intracellular iron stores (Embree et al. 2014). Future characterization of Gallionella and Geobacter could lead to the ability to tailor these metal-tolerant bacteria for bioremediation purposes.

4.3 Shewanella

Shewenella oneidensis MSR-1 is both a facultative anaerobe and a dissimilatory metal-reducing bacterium that inhabits aquatic environments and iron rich-sediments, and can use oxidized iron as a terminal electron acceptor (Bennett et al. 2015). As respiration occurs, Feo-transported Fe²⁺ accumulates inside of the bacterium and becomes incorporated into iron-utilizing proteins and solid-phase minerals (Bennett et al. 2015). In addition to Feo, Shewanella may possess another uncommon Fe²⁺ import system. The MgtE protein is a Mg²⁺/Co²⁺ transporter first identified in Bacillus firmus (Smith et al. 1995) but present in S. oneidensis MR-1, which contains three mgtE homologs. One of these homologs (SO_3966) is found in 26 of 36 Shewanella genomes and has been named the ferrous iron and cobalt importer (FicI) (Bennett et al. 2018). A role in Fe²⁺ uptake was proposed upon observations that a ΔficI strain of Shewanella conferred resistance to Fe²⁺. This hypothesis was confirmed by measuring Fe²⁺ uptake in a ΔficI strain which was ≈90% lower than that of the WT strain and also displayed Co²⁺ and Mg²⁺ sensitivity (Bennett et al. 2018). However, it seems that the FicI system is less important than the Feo system when iron is scarce, as the ΔfeoB strain was unable to grow under iron-limiting conditions unlike the ΔficI strain. It is suggested that FeoB serves as the primary Fe²⁺ transporter for Shewanella while FicI serves as a secondary Fe²⁺ transporter.

Due to its high rate of Fe²⁺ acquisition, Shewenella has had to adapt mechanisms to prevent iron overload (Bennett et al. 2015; Carlson et al. 2012; Dunning et al. 1998). One such mechanism is through the use of a ferrous iron exporter encoded by the gene locus SO_4475 and named FeoE (herein referred to as SoFeoE), although it is not associated with the feo operon. FeoE is predicted to be a member of the cation diffusion facilitator (CDF) family, which is a family of inner membrane proteins that efflux divalent metal ions using the proton motive force (Bennett et al. 2015; Nies and Silver 1995; Paulsen and Saier 1997). SoFeoE shares sequence similarity and identity (≈60% and ≈48%, respectively) with the YiiP (FieF) protein from E. coli, which has been shown to export Zn²⁺, Cd²⁺ and Fe²⁺ (Chao and Fu 2004; Wei and Fu 2005; Grass et al. 2005b). SoFeoE was also previously shown to export Zn²⁺ and Cd²⁺, but had not been tested for its ability to export Fe²⁺, either in vitro or in vivo (Coudray et al. 2013). One study examined the ability of SoFeoE to protect S. oneidensis MSR-1 against Fe²⁺ toxicity when utilizing ferric citrate for anaerobic respiration by examining the survivability of ΔfeoE strains. S. oneidensis deficient for feoE exhibited a growth defect compared to the WT strain but could be rescued with complementation by feoE (Bennett et al. 2015). Ferrozine assays measuring Fe²⁺ directly resulting from ferric citrate respiration were similar for both deletion and WT strains, supporting the hypothesis that respiration on ferric citrate is not impaired in the ΔfeoE strain, but functions to prevent Fe²⁺ toxicity (Bennett et al. 2015). This assertion was further supported by lower growth rates of the ΔfeoE strain when grown in the presence of 1 mM FeCl₂, and by iron retention assays demonstrating that feoE deficient S. oneidensis retain significantly more iron than the WT S. oneidensis (Bennett et al. 2015). Sensitivity of the ΔfeoE strain was tested against a range of divalent metal ions and feoE was found to be specific for Fe²⁺, in contrast with the previous suggestion of metal promiscuity (Bennett et al. 2015). Furthermore, the FeoE protein is conserved amongst Shewanella spp. suggesting that it is a common mechanism to protect against Fe²⁺ toxicity. Homologs of FeoE may also be present in Geobacter metallireducens and Geobacter sulfurreducens, but share a much lower sequence similarity (Bennett et al. 2015).

Ferrous iron uptake likely accounts for one of the dominant iron sources in this bacterial family, as Shewanella encode for just one natural siderophore, putrebactin (Pub), synthesized by the pubABC operon using putrescine as a precursor. The TonB-specific ferric siderophore receptor is encoded by putA and the Pub reductase is encoded by putB (Ledyard and Butler 1997; Kadi et al. 2008; Soe and Codd 2014; Liu et al. 2018). A spontaneous mutant of S. oneidensis, SO-X2, was identified and believed to be deficient for cytochrome c production as colonies of S. oneidensis appeared white instead of red (Dong et al. 2017). However, this was not the result of a mutation within the cytochrome c biosynthesis genes, but rather due to a mutation in the putA gene (Dong et al. 2017). The deletion of putA resulted in an ≈50% decrease in intracellular iron concentration when compared to the WT and also resulted in colorless colonies as the inability of the ΔputA strain to deliver iron to the cytosol has downstream effects on Shewanella’s ability to produce both cytochrome c and heme (Dong et al. 2017; Liu et al. 2018). Correlating with these results was the observation that the ΔputA strain also produced Pub at a significantly higher level under iron-limiting conditions compared to the WT (Dong et al. 2017). Knockouts of the pub operon and putB were also tested for their effects on iron uptake in Shewanella. Both Δpub and ΔputB strains exhibited normal levels of intracellular iron, though the Δpub strain was incapable of producing Pub. This result indicates the presence of a primary functional iron uptake system, such as Feo, that is capable to circumventing this loss (Liu et al. 2018). In this study, a Δfeo deletion strain had a lower total iron concentration than the WT, but also had significantly more total iron than the ΔputA strain. However, loss of feo appeared to be more detrimental to Shewanella growth when compared to the growth of the ΔputA strain. A double deletion strain grew poorly under iron depleted conditions, but could grow at Fe²⁺ concentrations of 0.5 mM and higher, likely using FicI as a secondary Fe²⁺ transporter (Liu et al. 2018). Thus, it is clear that ferrous iron uptake contributes strongly to iron homeostasis in Shewanella, and it is likely that similar mechanisms may be operative in other FeOB and FeRB, although more characterization is warranted.

5 The Host-Pathogen Interface

Among other functions, humans depend on mononuclear and dinuclear iron proteins for carnitine biosynthesis, hypoxic sensing, DNA biosynthesis, and functionalization of unsaturated lipids. Additionally, cellular iron is also incorporated into [Fe-S] clusters and utilized in heme biosynthesis, both of which serve as cofactors for enzymes involved in gene regulation, oxygen transport, and even drug metabolism (Hider and Kong 2013; Ganz 2008). The labile iron pool is composed mostly (>80%) of Fe²⁺, with iron concentrations estimated to be between ca. 10⁻⁷ M and ca. 10⁻⁶ M in either erythroid cells or hepatocytes (Hider and Kong 2013). By way of comparison, one estimate places free iron concentrations in human serum at ≈10⁻²⁴ M (Fischbach et al. 2006). Because ferrous iron is significantly more soluble and more kinetically labile compared to ferric iron at physiological pH (≈7.4), ferrous iron is typically the species that is transported into cells, translocated within cells, and even incorporated into iron-dependent enzymes and proteins (Hider and Kong 2013). Thus, the prevalence of Fe²⁺ in human enzymes and within the labile iron pool suggests that invading microbes might encounter a significant amount of Fe²⁺ within their colonization niches.

The concentration of iron in the labile Fe²⁺ pool is governed by several factors. The amount of iron required for normal physiology may differ among both organisms and cells (e.g., mitochondria have a higher iron requirement since that is where heme and iron-sulfur cluster biosyntheses take place). One mechanism for regulating the intracellular labile iron pool is to control the rate of iron transport into cells by divalent metal ion transporters (Hider and Kong 2013). Iron within the labile iron pool can be temporarily depleted be oxidation of iron and storage within ferritins. Finally, iron efflux via ferroportin can also control the concentration of the labile iron pool, preventing iron overload and potential redox stress within the cell (Hider and Kong 2013). Together, these mechanisms maintain iron homeostasis in humans, and these systems are tightly regulated to control iron availability during times of infection.

Iron homeostasis in humans and animals is heavily altered during microbial infection. One of the host’s first defense mechanisms against invading pathogens is to reduce the absorption of dietary iron. The host also sequesters iron already present in the extracellular space to further limit the availability of this essential nutrient to pathogens. These processes combined are often termed “iron withholding” and constitute a part of the host defense mechanism known as nutritional immunity (Ganz 2008; Abu Kwaik and Bumann 2013; Skaar 2010). Bacteria are estimated to require cytoplasmic iron pools at a concentration of ≈10⁻⁶ M, and the host can significantly limit the amount of iron available to microbes depending on the niche in which they colonize. In response, bacteria must overcome these limiting conditions in order to survive (Fischbach et al. 2006). Some pathogens have adapted to live within the cytosol of eukaryotic cells where essential nutrients, including iron, are plentiful. Bacteria may also secrete proteases that degrade host proteins and enzymes, releasing essential nutrients that may be acquired by pathogens (Parrow et al. 2013). One important mechanism utilized by bacteria to establish infection is Fe²⁺ uptake driven predominantly by the Feo system. This section summarizes our knowledge of ferrous iron acquisition during infection of unicellular pathogens, which is strongly linked to the Feo system.

5.1 Cellular Studies of Ferrous Iron Acquisition at the Host-Pathogen Interface

Helicobacter pylori is the causative agent of chronic gastritis and though several virulence factors had been identified in H. pylori, its mechanisms of iron transport and homeostasis are not well understood. H. pylori does not appear to produce siderophores but relies on both Fe²⁺ and heme uptake as well as iron sequestered from human lactoferrin. H. pylori also possesses ferric reductase activity, which likely facilitates Fe²⁺ transport via the Feo system (Velayudhan et al. 2000). Inactivation of feoB in H. pylori significantly reduced the rate of ⁵⁵Fe uptake regardless of oxidation state, and this phenotype could be restored by complementation with feoB. Fe²⁺ uptake kinetics have revealed that FeoB constitutes a high affinity Fe²⁺ transporter (apparent Michaelis constant of ca. 0.5 μM), and the presence of a second, unidentified, low affinity Fe²⁺ transporter was also suggested. Though FeoB is considered to be specific for Fe²⁺ transport, the presence of Cu²⁺ in a 100-fold excess of Fe²⁺ inhibited Fe²⁺ uptake by 74%, while Co²⁺, Mn²⁺, Ni²⁺, and Zn²⁺ had no inhibitory effect on Fe²⁺ transport (Velayudhan et al. 2000). Intriguingly, the ATP synthesis inhibitor, DCCD, strongly inhibited Fe²⁺ transport, perhaps due to NTP promiscuity of this organism’s FeoB. Inhibition also occurred in the presence of orthovanadate, a molecule that inhibits ATP hydrolysis. These results suggested that FeoB-mediated Fe²⁺ transport could utilize active ATP hydrolysis as an energy source for the transport process (Velayudhan et al. 2000).

Legionella pneumophila colonizes human macrophages and freshwater amoebae within a parasitic vacuole, where Fe²⁺ likely dominates (Robey and Cianciotto 2002). An feoB mutant transported virtually no radiolabeled Fe²⁺ but did not exhibit complete attenuation for extracellular growth in iron-limited media. However, the feoB mutant also confers increased resistance to the antibiotic streptonigrin, a direct result of decreased iron pools in the mutant as streptonigrin toxicity is iron-dependent (Robey and Cianciotto 2002). Upon co-culturing the feoB mutant with the amoeba Hartmannella vermiformis under iron-limiting conditions, the mutant strain was detected at a significantly diminished concentration, indicating that FeoB is necessary for growth and survival in L. pneumophila (Robey and Cianciotto 2002). The role of FeoB for intracellular colonization and survival was confirmed in a human macrophage model where the mutant L. pneumophila was attenuated in its ability to kill macrophages. Furthermore, a mouse model of infection demonstrated that the lungs of the mice contained lower concentrations of the feoB mutant compared to the WT, confirming the importance of feoB to L. pneumophila virulence (Robey and Cianciotto 2002).

Campylobacter jejuni is a causative agent of foodborne gastroenteritis, and the connection between FeoB and C. jejuni pathogenesis was first investigated in 2003 (Raphael and Joens 2003). Iron uptake and growth of different C. jejuni strains were compared to that of E. coli W3110. The acquisition of ⁵⁵Fe²⁺ in C. jejuni was generally lower than that of E. coli W3110 (Raphael and Joens 2003). An insertion mutation of feoB in C. jejuni was then analyzed for ⁵⁵Fe²⁺ uptake and rates were similar to those observed in the WT strains. The experiments were repeated with a feoB mutation strain generated through allelic exchange, but no difference in ⁵⁵Fe²⁺ uptake was observed, suggesting ⁵⁵Fe²⁺ uptake was not mediated by FeoB and that FeoB was not be required for Fe²⁺ uptake in C. jejuni (Raphael and Joens 2003). However, a later study reinvestigated the role of FeoB in C. jejuni as the genomes for the strains previously investigated had not been sequenced, and the presence of an additional FeoB homolog could not be precluded (Naikare et al. 2006). In C. jejuni NCTC 11168, feoA and feoB were found to be cotranscribed and thus constituted an operon. Interestingly, feoA expression was elevated six to eight-fold over that of the WT strain in feoB mutants (Naikare et al. 2006). Furthermore, the feoB mutant acquired Fe²⁺ at ten-fold less compared to the WT strain. Isolation of spheroplasts from the WT and ΔfeoB strain showed that 90% of the transported iron was contained within the cytosol, suggesting that C. jejuni can also transport Fe²⁺ via a different transporter (Naikare et al. 2006). FeoB also promoted the growth of C. jejuni in iron limited media compared to the ΔfeoB strain. Analysis of iron content during these experiments revealed that a majority of accumulated ⁵⁵Fe²⁺ in the ΔfeoB strain was localized in the periplasm suggesting the lack of a functional inner membrane transporter (Naikare et al. 2006). C. jejuni 81-176 and a feoB mutant were then tested for their ability to invade human INT-407 embryonic intestinal cells and porcine IPEC-1 small intestinal epithelial cells. Both strains were equally able to invade both types of cells and survive for up to 48 h. However, after 48 h the feoB mutant exhibited a decreased ability to survive in the eukaryotic cells indicating that feoB contributes to intracellular survival in C. jejuni (Naikare et al. 2006). The feoB mutants examined in this study also exhibited a decreased ability to colonize in a chick cecum model and in a rabbit ileal loop model demonstrating that FeoB is important for colonization within the host (Naikare et al. 2006). Lastly, C. jejuni WT strains and feoB mutants were examined for their ability to colonize in a piglet intestine to simulate infection and colonization in a human host. Three days post-infection, different parts of the gastrointestinal tract of the piglets were harvested and through competition assays it was determined that feoB mutants were unable to colonize in the intestinal tract of the piglets (Naikare et al. 2006).

Francisella tularensis is responsible for zoonotic tularemia infections, is highly infectious, and has a high mortality rate, which can be linked in part to Feo utilization (Thomas-Charles et al. 2013). F. tularensis ΔfeoB strains exhibit reduced colony size as a result of inefficient Fe²⁺ uptake, and this strain exhibited significantly slower growth in iron-restricted medium compared to the WT strain. The loss of FeoB also results in increased secretion of the rhizoferrin-like siderophore encoded by the fsl operon suggesting that F. tularensis attempts to overcome the loss of FeoB through an alternative iron uptake system (Thomas-Charles et al. 2013). The ΔfeoB strain also contained significantly reduced cellular iron content, which could be restored with complementation. Unsurprisingly, the strain displayed significantly reduced replication in human lung epithelial cells and macrophages. However, despite the lack of a functional feoB, the deletion strain still caused 100% mortality rate in infected mice; however, at 3 days post-infection, the lungs, livers and spleens were collected to analyze the presence of each strain and it was determined that the ΔfeoB strain was present in reduced quantities (Thomas-Charles et al. 2013). Another study of F. tularensis Schu4 demonstrated that FeoB was similarly necessary for bacterial growth, Fe²⁺ uptake under both iron-replete and iron-deplete conditions, and that in a ΔfeoB’ (encoding FeoB without the C-terminus) strain the fsl operon displayed increased expression (Pérez et al. 2016). The ΔfeoB’ strain could replicate within murine macrophages but showed reduced replication in human liver carcinoma cells. A double deletion strain also lacking the fslA gene could not replicate within the murine macrophages or the human liver carcinoma cells. A murine model demonstrated that the ΔfeoB’ strain could kill infected mice within 5–7 days, similar to the WT F. tularensis Schu4, while mice infected with the double deletion strain lived for at least 21 days post-infection, demonstrating the importance of FeoB coupled to other iron acquisition systems for full bacterial virulence (Pérez et al. 2016).

In a similar manner, the Feo system of the avian pathogen E. coli O78 (APEC) is connected to other metal uptake routes for maximum virulence (Sabri et al. 2008). In APEC, SitABCD contributes to virulence in chickens, and strains lacking this system exhibited significantly reduced colonization. However, in the presence of a functional SitABCD system, strains carrying either a ΔfeoB or a ΔmntH behaved similar to WT APEC and persisted in the chicken model, suggesting Feo and MntH are less important for APEC virulence compared to SitABCD (Sabri et al. 2008). Double deletion strains were tested for their effects on APEC virulence as well. Both a ΔsitABCDΔmntH strain and a ΔsitABCDΔfeoB strain were just as attenuated for virulence as the ΔsitABCD strain, confirming the previous result that feoB and mntH do not contribute to APEC virulence as much as sitABCD (Sabri et al. 2008). Interestingly, the ΔsitABCDΔfeoB strain was found at lower levels in the blood, lungs, and spleen of the chicken model but was present in higher levels in the liver. These results suggest that each metal transport system may be important for colonization in certain niches within the host (Sabri et al. 2008).

Shigella flexneri infections cause dysentery in humans, and this pathogen also utilizes multiple divalent metal ion uptake systems including Feo to maintain proper metal homeostasis. Akin to APEC, a feoB mutant strain was modestly attenuated for growth, similar to an iucD mutant, which is a component of the aerobactin siderophore synthesis system. Interestingly, a double mutant of feoB and iucD grew poorly, though it is not understood why (Runyen-Janecky et al. 2003). Henle cells were utilized to determine whether the mutant S. flexneri could form plaques. Single mutants of sitA, feoB, and iucD all formed plaques comparable to WT S. flexneri. However, the loss of any two of the three genes resulted in reduced plaque formation and size, suggesting a link between Fe²⁺ uptake and plaque formation. Lastly, a triple mutant deficient for all three genes was fully attenuated for plaque formation (Runyen-Janecky et al. 2003). These data suggest that each Fe²⁺ transport system may be important for growth, virulence, and/or survival of S. flexneri.

Yersinia pestis, the pathogen responsible for bubonic, septicemic, and pneumonic plagues in humans and animals, also utilizes the Feo system for virulence. Y. pestis is strongly adapted to scavenge iron from its environment as it encodes up to 12 putative iron transport systems, including Feo. The Yfe system, homologous to the Sit system, transports Mn²⁺ and Fe²⁺ and has been shown to be important for pathogenesis in mice (Perry et al. 2007). Y. pestis encodes for its own siderophore, Yersiniabactin (Fe³⁺-utilizing), which was previously shown to be an essential virulence factor during the early stages of infection, whereas the Yfe system (Fe²⁺-utilizing) was indispensable for late stage infection, delineating a clear role for Fe³⁺ and Fe²⁺ at different stages of infection (Perry et al. 2007). Under microaerophilic conditions, a Y. pestis ΔfeoB strain was attenuated for growth by ≈50%, nearly to the same extent as a Δyfe strain. The loss of both Fe²⁺ transport systems attenuated growth further. These effects were not observed in the deletion strains growing under oxic conditions (Perry et al. 2007). Interestingly, the FeoC protein does not appear to be essential for Y. pestis growth, while the loss of a functional FeoA attenuated growth to the same extent as the ΔfeoB strain. The Y. pestis ΔfeoB strain and the Y. pestis Δyfe strain were still able to replicate intracellularly in murine macrophages, whereas a strain lacking both feoB and yfe could not replicate in the murine macrophages, indicating that one system can compensate for the loss of the other during infection (Perry et al. 2007).

A 2009 study demonstrated the importance of FeoB to the virulence of Streptococcus suis, a pathogen that infects swine and is responsible for septicemia and meningitis. In silico analysis of S. suis led to the discovery of an feoA gene in S. suis predicted to encode a larger than usual 156 amino acid protein and feoB was identified and predicted to encode a 714 amino acid protein. No feoC gene was identified (Aranda et al. 2009). Using electrophoretic mobility shift assays (EMSAs), the S. suis Fur protein was found to bind specifically to the feo operon’s promoter, and derepression of feoA and feoB expression was observed in a fur mutant. A murine infection model was then used to establish the contribution of feo to virulence. In a feoB mutant strain, in vitro growth of the mutant was lower in iron-depleted conditions, and virulence in mice was significantly attenuated, linking feoB to full virulence of S. suis in murine models (Aranda et al. 2009).

P. aeruginosa is an opportunistic pathogen reliant on host iron that is obtained via multiple acquisition systems, and this pathogen is the dominant infectious agent of late-stage cystic fibrosis (CF) patients. P. aeruginosa’s reliance on host iron is evident, as transcripts of feoA, feoB, bqsR/S (a two-component Fe²⁺ sensing system), pvdA (the pyoverdine biosynthetic protein), fptA (the ferripyochelin receptor), and hasAp (the heme uptake protein) have all been detected at various levels in the sputum. As CF progresses and becomes more severe, O₂ permeability of lung tissue generally decreases, and the concentration of Fe²⁺ in lung sputum increases and is correlated with declining lung function (Hunter et al. 2013). One possible source of reduced iron could come from reaction with excreted reducing agents, such as phenazines, which are redox-active compounds that can reduce Fe³⁺ to Fe²⁺. Intriguingly, the phenazines pyocyanin and phenazine-1-carboxylic acid are present in ca. 80% of sputum samples; however, phenazine levels do not appear to correlate with increased Fe²⁺ concentrations (Hunter et al. 2013), making the source of Fe²⁺ in the CF lung unclear. Nevertheless, it has been suggested that iron in CF lungs exists in mixed oxidation states at various stages of disease progression, and that the iron composition changes over time (Hunter et al. 2013), likely an important factor for biofilms. There is also evidence that FeoB is necessary for P. aeruginosa to utilize Fe³⁺ derived from citrate transport across the outer membrane receptor FecA (Marshall et al. 2009). Thus Feo appears to play an important role in oxic iron transport, in addition to anoxic iron transport within P. aeruginosa.

CF disease progression is correlated with increased biofilm formation by P. aeruginosa, and iron availability is an important factor for biofilm formation. A biofilm assay under hypoxic conditions demonstrated that an Fe³⁺ chelator, conalbumin, and a Fe²⁺ chelator, ferrozine, could inhibit biofilm formation by >50%. Biofilm formation could be restored by the addition of excess iron, suggesting that iron bioavailability is important for biofilm formation in cystic fibrosis lungs (Hunter et al. 2013). Both chelators were then tested for their ability to dissolve preformed biofilms under oxic and anoxic conditions. Significant biofilm dissolution was observed when both chelators were present. Under anoxic conditions, ferrozine alone could dissolve preformed biofilms by 20%. Biofilms reformed with the addition of iron, suggesting biofilm formation is iron-mediated (Hunter et al. 2013). This interesting corollary between biofilms and iron availability could be a future direction of exploration into the progression of CF disease.

Acinetobacter baumannii has recently been listed as one of the most dangerous opportunistic pathogens by the World Health Organization, and Feo is one of the contributing factors to this organism’s virulence (Álvarez-Fraga et al. 2018). 50 clinical strains of A. baumannii containing the Feo operon (feoA and feoB) were previously identified, and it was found that overexpression of feoA occurred during infection of A. baumannii within the lung. A clinically-relevant strain lacking the feoA gene was constructed and growth of this strain was tested under both iron-sufficient and iron-limited conditions. In iron-replete media, both the ΔfeoA and WT strains showed no differences in growth; however, the generation time of the ΔfeoA strain was significantly higher in iron-limited media compared to the WT strain, indicating that feoA is important for the fitness of A. baumannii (Álvarez-Fraga et al. 2018). In experiments assessing biofilm formation and cellular adhesion, the inactivation of feoA detrimentally affected both markers of A. baumannii virulence. When the ΔfeoA strain was complemented with a plasmid encoding for feoA, both abilities were only partially restored. Cells lacking feoA were also more susceptible to oxidative stress as opposed to the WT strain (Álvarez-Fraga et al. 2018). A Galleria mellonella (moth) infection model and a murine pneumonia model have both been used to study how feoA affects the virulence of this pathogen. A. baumannii ΔfeoA strains were unable to infect and kill G. mellonella. Similarly, mice that were intratracheally infected with the ΔfeoA strain had significantly decreased mortality rates, and the occurrence of positive sterling blood cultures for mice infected with the ΔfeoA strain increased by 75% compared to the WT, demonstrating that feoA is necessary for A. baumannii virulence (Álvarez-Fraga et al. 2018). Thus, these results suggest that FeoA could serve as a novel therapeutic target given its strong contribution to A. baumannii virulence.

P. gingivalis is an oral pathogen and the causative agent of gingivitis, one of the most common human infections worldwide, and its genome contains an feoA-feoB fusion. P. gingivalis has a growth requirement for iron, but it cannot synthesize heme, does not produce siderophores, and lacks ferric iron reductase activity, necessitating Fe²⁺ uptake as one of its major routes of iron acquisition (Dashper et al. 2005). At first glance, two genes annotated as feoB are present in the organism (feoB1 and feoB2). The feoB1 gene was demonstrated to encode a ferrous iron transporter necessary for pathogenesis (lesion formation) in murine models, while feoB2 was demonstrated to encode a functional manganese transporter. Interestingly, in the absence of the functional ferrous iron transporter (feoB1), the functional manganese transporter (feoB2) may be upregulated to support growth and survival, but this gene product does not support infection. This is the first example of an FeoB-like protein transporting a metal other than ferrous iron (Dashper et al. 2005), but it remains unclear whether feoB2 is a bona fide FeoB protein. In contrast, it is clear that the feoB1 gene encodes for a naturally-tethered FeoA-FeoB polypeptide based on sequence conservation. Beyond this information, no work at the protein level has detailed the function of this novel fusion.

B. fragilis is a commensal bacterium that can become opportunistic when and if it enters environments outside of the gut, is becoming increasingly resistant to metronidazole (the most widely used treatment for B. fragilis infections), and its genome also contains an feoA-feoB fusion (Veeranagouda et al. 2014; Rocha et al. 2019). Intriguingly, the absence of the feoAB gene conferred metronidazole resistance to B. fragilis in one study, despite exhibiting slower growth kinetics. However, the reason for increased resistance to metronidazole in feoAB mutants and the effect of iron on resistance remains unknown (Rocha et al. 2019). Additionally, both heme and non-heme iron are necessary for the survival of B. fragilis during infection (Rocha et al. 2019). B. fragilis cannot synthesize heme de novo, but it has an ability to remove metal from nonferrous metallated porphyrins (dechelatase activity) and to then insert ferrous iron into the apo porphyrins (ferrochelatase activity) (Rocha et al. 2019). The acquisition of ferrous iron is thus important to heme metabolism in B. fragilis, and a ΔfeoAB strain was unable to grow in the presence of heme, suggesting that iron derived from heme is necessary for growth (Rocha et al. 2019). Consistent with this hypothesis, when supplemented with 100 μM FeSO₄ growth of the deletion strain was rescued. It is suggested that iron is removed from heme in the periplasm and transported via FeoAB to support growth, thus linking heme metabolism and Feo-mediated Fe²⁺ uptake in this organism (Rocha et al. 2019). However, like P. gingivalis, work at the protein level on the B. fragilis FeoA-FeoB fusion is unrealized.

Finally, pathogens utilizing the Feo system also infect plants, although this host-pathogen interface is not well-studied. Xanthomonas is a genus of Gram-negative bacteria responsible for causing approximately 400 different diseases in plants, including bacterial blight disease in rice (Pandey and Sonti 2010). Xanthomonas oryzae pv. oryzae causes disease in the xylem vessels of rice leaves, and this pathogen encodes for a tripartite FeoA, FeoB, and FeoC system. These genes are cotranscribed when iron is both replete and limiting (Pandey and Sonti 2010). In X. oryzae, an feoB mutant was defective for growth and overproduced siderophores, while the WT strain did not, suggesting that the pathogen was trying to overcome the loss of an essential iron transport system (Pandey and Sonti 2010). Rice inoculated with the Xanthomonas feoB mutant produced fewer lesions than those inoculated with WT, whereas bacteria deficient for siderophore production were as virulent as WT, suggesting that the Feo system is an essential component of in planta virulence (Pandey and Sonti 2010).

6 Conclusions and Outlook

From early to present-day Earth, multiple pools of ferrous iron are important for microorganisms to meet their metabolic needs. This principle is especially true in the anoxic and acidic environments that bacteria encounter within a host or in soils, aquatic environments, or waste sites, to name a few. Given the drastic changes in environments that microbes encounter, from iron-replete to iron-depleted conditions, it is absolutely essential for iron uptake and storage to be regulated in a manner that maintains intracellular iron stores at appropriate levels and fulfills the varying metabolic needs of each organism.

Several Fe²⁺ uptake systems have been identified in bacteria, though many are not specific for Fe²⁺ and have not been well-characterized either at the cellular level or at the protein level, and thus their contribution to bacterial virulence remains unknown. In contrast, cellular studies have established the role of Feo (and in particular FeoB), the dominant prokaryotic Fe²⁺ transport system, for bacterial survival and virulence. Despite this importance, several gaps in our knowledge still remain regarding the structure of full-length FeoB and how FeoB facilitates transport of Fe²⁺ across the cytoplasmic membrane. It is unclear if a periplasmic binding protein or another chelator facilitates transfer of Fe²⁺ to FeoB, potentially through the periplasmic loop, which residues in the TM region binding and translocating Fe²⁺ across the membrane, and what happens to Fe²⁺ once it reaches the cytosol. Additional questions remain regarding the GTPase/NTPase activity of the G-protein domain of NFeoB and how nucleotide hydrolysis is linked to Fe²⁺ transport. FeoA and FeoC, both small cytosolic proteins, have been subjected to numerous studies though their functions remain undetermined. Given that FeoA can be found fused to NFeoB through a linker region to the G-protein domain, it is likely that FeoA and NFeoB interact. However, the exact site of this potential interaction and the functional implications have not been elucidated. Lastly, FeoC has been shown to bind a redox active [4Fe-4S] cluster, implying that FeoC could add an additional layer of regulation to the Feo system. However, FeoC is poorly conserved and some FeoC proteins, such as from Vibrio species, do not contain the conserved Cys residues required for iron-sulfur cluster binding. Undoubtedly, more experiments aimed at elucidating these details and more functional models of FeoB will be necessary to gain a better understanding of this broadly distributed Fe²⁺ transport pathway.

Despite this lack in knowledge, it is clear that ferrous iron acquisition, linked in particular to the Feo system, is important for the survival of bacterial species across multiple environments, and this point is especially true of pathogenic bacteria that are becoming alarmingly problematic. Antibiotic resistance has become one of the world’s greatest public health challenges. Both the challenges and breakthroughs associated with antibiotic resistance continue to change as resistant pathogens evolve. According to a 2019 report from the Centers for Disease Control and Prevention (CDC), there are 2.8 million antibiotic-resistant infections in the United States every year which results in more than 35,000 deaths (Antibiotic resistance threats in the United States 2019). In fact, in the United States it is estimated that deaths associated with resistant infections will outpace those associated with cancer by the year 2050. As iron acquisition is linked to pathogenesis, and bacteria lacking functional ferrous iron transport systems (such as Feo) have been shown to be avirulent or unable to colonize in a host-associated environment, a greater understanding of the mechanisms underlying ferrous iron acquisition could be leveraged for the development of new therapeutics that could help to combat the growing global emergency of antibiotic resistance.