Diversity of Structures and Functions of Oxo-bridged Non-heme Diiron Proteins

Abstract

Oxo-bridged diiron proteins are a distinct class of non-heme iron proteins. Their active sites are composed of two irons that are coordinated by amino acid side chains, and a bridging oxygen that interacts with each iron. These proteins are members of the ferritin superfamily and share the structural feature of a four α-helix bundle that provides the residues that coordinate the irons. The different proteins also display a wide range of structures and functions. A prototype of this family is hemerythrin, which functions as an oxygen transporter. Several other hemerythrin-like proteins have been described with a diversity of functions including oxygen and iron sensing, and catalytic activities. Rubrerythrins react with hydrogen peroxide and rubrerythrin-like proteins possess a rubredoxin domain, in addition to the oxo-bridged diiron center. Other redox enzymes with oxo-bridged irons include flavodiiron proteins that act as O₂ or NO reductases, ribonucleotide reductase and methane monooxygenase. Ferritins have an oxo-bridged diiron in the ferroxidase center of the protein, which plays a role in the iron storage function of these proteins. There are also bacterial ferritins that exhibit catalytic activities. The structures and functions of this broad class of oxo-bridged diiron proteins are described and compared in this review.

1. Introduction

It has been estimated that as many as half of all proteins contain a metal [1, 2]. Iron is a common component of metalloproteins that allows the protein to perform a variety of functions. Iron is a component of the heme cofactor that is used to store and transport oxygen as in hemoglobin. Iron in c-type hemes typically functions in electron transfer through and between proteins. Iron in b-type hemes can be used to activate oxygen during catalysis, as in oxidases and oxygenases. Iron-sulfur centers also participate in electron transfer reactions. Non-heme iron is utilized by proteins for catalysis and electron transfer.

This review focusses on a functionally diverse and expanding class of metalloproteins that have two non-heme iron ions with a bridging oxygen atom. The chemical state of the bridging oxygen is typically O, which may be protonated or hydrogen bonded depending upon the environment in the protein and redox state of the irons. However, in this diverse group of proteins other states, such as H₂O and peroxy are observed. It is generally believed that the assembly of the oxo-bridged diiron sites do not require accessory proteins but are self-assembled. A prototype of the oxo-bridged diiron protein is hemerythrin [3, 4]. The source of the bridging oxygen is water [3]. True hemerythrins bind O₂ at one of the irons and function as oxygen transporters. Several other types of hemerythrin-like proteins have been described with other functions that include iron or O₂ sensing, iron-sulfur center repair, NO reductase activity and catalase activity [5, 6]. Rubrerythrins use the oxo-bridged diiron for oxidative stress tolerance [7]. Other identified enzymes that utilize an oxo-bridged diiron at their active sites include flavodiiron proteins that reduce NO or O₂ [8], ribonucleotide reductase [9] and methane monooxygenase [10]. All of these oxo-bridged diiron proteins are considered part of the ferritin superfamily. Ferritins have oxo-bridged diiron centers that can function in iron storage, prevention of oxidative damage to nucleic acids and catalase activity [11, 12]. All of these oxo-bridged diiron proteins share a common structural fold, a 4 α-helix bundle that houses the diiron site. However, the overall structures can be quite different and the identity of the ligands that coordinate the irons varies. Oxo-bridged diiron proteins are found in prokaryotes, eukaryotes and archaea. Many of the members of this broad group of proteins have been reviewed individually. This review compares and contrasts the structures of the diiron sites and overall structures of these proteins, and highlights the diversity of their functions.

2. Hemerythrin and Hemerythrin-like proteins (HLPs)

2.1. True hemerythrins

The first protein to be structurally characterized and shown to possess an oxo-bridged diiron center is hemerythrin [13]. True hemerythrins were first described in marine invertebrates, and subsequently numerous hemerythrin-like proteins (HLPs) that contain an analogous oxo-bridged diiron center were described in a wide range of prokaryotic and eukaryotic species [5]. Hemerythrins do not contain heme, but use oxo-bridged non-heme irons for oxygen transport or storage [3]. Hemerythrins have a core structure composed of a four right-handed α-helix bundle fold that houses the diiron site with the bridging oxygen. These proteins are typically comprised of six or eight identical subunits, as observed in the Phascolopsis gouldii crystal structure [14] shown on Figure 1A. However, monomeric hemerythrins have also been characterized from Methylococcus capsulatus (Bath) [15] (Fig. 1B), and the peanut worm Themiste hennahi [16]. In hemerythrins, the two oxo-bridged irons are coordinated by seven amino acid residues: five His, one Glu and one Asp, as shown in Figure 1C.

Figure 1. A.

A. Structure of the octomeric hemerythrin from Phascolopsis gouldii (PDB entry 1I4Y) [14]. B. Structure of the canonical four 〈-helix bundle of the monomeric hemerythrin from M. capsulatus (Bath) (PDB entry 4XPX) [15]. The two irons shown as brown spheres. C. The oxo-bridged diiron center of M. capsulatus hemerythrin (PDB entry 4XPX) [15]. Amino acid residues that coordinate irons (brown spheres) are shown as sticks, and the bridging oxygen is shown as a red sphere. Figures were prepared using PyMOL [17].

Spectroscopic and redox studies of the monomeric hemerythrin from M. capsulatus (Bath) characterized three different oxidation states of the diiron center (Figure 2). The oxy form is Fe⁺³-Fe⁺³ with O₂ bound to iron. The terminal O typically reacts with a hydrogen and becomes a hydroperoxyl ligand that can hydrogen bond with the bridging oxygen. The deoxy form is Fe⁺²-Fe⁺². The oxy and deoxy forms can directly interconvert by dioxygen release or incorporation. The oxy form may also auto-oxidize in the presence of water to yield the met form, with release of H₂O₂ [15, 18]. This met form is Fe⁺³-Fe⁺³ without bound O₂, and it can be reduced to regenerate the deoxy form. The met form can bind ligands other than O₂, such as OH, or unnatural ligands such as azide [4, 13].

Figure 2.

Interconversion of the deoxy, oxy, and met forms of the oxo-bridged diiron center of hemerythrin.

2.2. Hemerythrin-like proteins (HLPs)

Bioinformatics analyses have identified a large number of HLPs in all phylogenetic kingdoms [19]. All HLPs possess oxo-bridged diiron sites within a core structure of four α-helices. The amino acid residues that ligate the two irons in HLPs are similar but with some variation. The sequence of each exhibits the characteristic HHE cation-binding domain seen in true hemerythrins. The residues that provide iron ligands in hemerythrins are in the sequence in the order HHEHHHD, with variable lengths of separation between residues in the sequence motif [3]. Some HLPs possess this same sequence pattern; however, others exhibit HHEYHHE, HHEHHE or HHEEHHE patterns [5, 6] (Figure 3).

Figure 3.

Examples of the different combinations of amino acid residues that coordinate the irons in hemerythrin-like proteins. A. HLP from Mycobacterium kansasii with 4H/2E/1Y coordination (PDB entry 6Q09) [20]. B. Iron-sulfur center repair HLP, YtfE, from E. coli with 4H/2E coordination (PDB entry 5FNN) [21]. C. Iron sensing FBXL5 HLP from Homo sapiens with 4H/3E coordination (PDB entry 3V5X) [22]. Figures were prepared using PyMOL [17].

The HHEYHHE motif is seen only in HLPs from mycobacterial species [6]. The mycobacterial HLPs are also distinct in possessing a fifth α-helix connected by a short loop to the four-helix core structure that houses the oxo-bridged diiron site. The protein from Mycobacterium tuberculosis is a homodimer [23] and that from Mycobacterium kansasii is a monomer [20]. The M. tuberculosis HLP exhibits catalase activity [23] and the M. kansasii HLP exhibits modest catalase activity and reactivity towards nitric oxide [20].

The HLPs that have been identified from all other sources have very different overall structures. In each, the four α-helical domain is fused to another protein domain that dictates the biological function of the HLP. These functions include signal transduction [24], oxygen sensing [25], iron sensing [26], repair of Fe-S centers from redox damage [21, 27], modulation of the oxidative stress response [28], and storage and transport of iron [29]. HLPs have also been linked to antimicrobial [30] and antiviral activity [31], and immune response in some organisms [32]. The function of many HLPs is linked to other proteins domains of known function. Some examples are as follows. The F-BoX and Leucine-rich repeat 5 (FBXL5) protein domain is associated with some HLPs that function as iron sensors, in which the oxo-bridged diiron structure destabilizes under iron depletion [33]. The F-BoX domain is part of the E3 ubiquitin ligase complex that targets iron regulated protein 2 (IRP2) for degradation, which is delayed under hypoxic conditions [34]. The FBXL5 domain of an HLP also binds the Rho GDP inhibitor 2 (RhoGD2) and attenuates the resistance for the antineoplastic drug cisplatin, and therefore is a potential target to treat this gastric cancer cell resistance [35]. An HLP was described with an additional domain that has cyclic-di-GMP phosphodiesterase activity [28] that is suggested to be an oxygen and redox sensor that modulates cellular GMP levels [28]. An HLP was described that has iron-dependent DNAse activity and is possibly related to DNA damage induced by oxidative stress [36].

3. Rubrerythrin and rubrerythrin-like proteins

The name rubrerythrin stems from the fact that canonical rubrerythrins possess a C-terminal rubredoxin domain and N-terminal domain that is similar to that of hemerythrin, with an oxo-bridged diiron center (Figure 4A). The rubredoxin domain is smaller than the hemerythrin domain. It is similar to the rubredoxin as it contains an Fe-4Cys center, in which an iron is coordinated by four sulfur ligands provided by the four Cys residues [37] (Figure 4B). The oxo-bridged diiron center in the hemerythrin domain is coordinated by Glu residues and a single His residue (Figure 4C). In each domain, the irons are ferrous under physiological conditions [38].

Figure 4.

Structural features of the rubrerythrin from D. vulgaris. A. The rubrerythrin monomer with the hemerythrin-like domain is pale cyan and the rubredoxin-like domain is light blue. B. The Fe-4Cys rubredoxin-like site. C. The hemerythrin-like diiron center. The Fe, N, O and S atoms are colored brown, blue, red and yellow, respectively. The figures were prepared with PyMOL [17] using PDB entry 1RYT [37].

The rubrerythrin isolated from D. vulgaris is a dimer in solution [39]. The hemerythrin-like domains of rubrerythrins possess a twisted left-hand 4-helix bundle fold (Figure 4A). As shown in the structure of the Pyrococcus furiosus rubrerythrin (Figure 5), the protein is a head-to-tail homodimer. What this means is that the rubredoxin-like domain from one monomer (“head’) interacts with the hemerythrin-like domain on the other monomer (“tail”). In the hemerythrin-like domain, each monomer contributes two helices to form the 4-helix bundle that surrounds the diiron site [37, 40] (Figure 5). In contrast to true hemerythrins, the hemerythrin-like domains of rubrerythrins have greater solvent accessibility and do not use O₂ as an oxidant [37, 41].

Figure 5.

Dimeric structure of rubrerythrin from P. furiosus. Each monomer is colored gray or cyan. Irons are shown as brown spheres. The figure was prepared with PyMOL [17] using PDB entry 3PWF [40].

Rubrerythrins are scavengers of H₂O₂ that play a role in oxidative stress defense. Rubrerythrin homologues are found in protozoa, bacteria [42] and archaea [40]. The proteins are classified in five groups, in part on their evolutionary relationship. Groups 1, 2 and 3 proteins are present in obligate anaerobes. Group 4 proteins are present in facultative aerobes. Group 5 proteins are present in cyanobacteria [7]. Most of the canonical rubrerythrins belong to Group 1. Many of those in Group 2 are termed “reverse-type rubrerythrins” because the order of their rubredoxin and hemerythrin domains is reversed. In other words, these proteins have an N-terminal rubredoxin domain and C-terminal hemerythrin domain. All members of group 4, and about 15% and 40% percent of the groups 2 and 5, respectively, lack their rubredoxin-like domain.

The substrate for rubrerythrins is H₂O₂. In many anaerobic hosts the H₂O₂ is generated by superoxide reductases [43]. Rubrerythrin converts the H₂O₂ to water by a two-electron reduction [44]. This is an important mechanism in microbes that do not express superoxide dismutases to prevent oxidative damage [45]. A reaction mechanism for rubrerythrin has been proposed that is based on structural differences between the oxidized, reduced, and adduct-intermediate forms of the enzyme. The rubredoxin-like domain obtains electrons from an external reductant. For some rubrerythrins, the electron donor is NAD(P)H [44, 46] (Eq. 1). This is followed by two one-electron transfers from the Fe-4Cys center to the diferric diiron center. The resultant diferrous center then binds H₂O₂ which bridges the two irons to form a μ−1,2-H₂O₂ intermediate that is converted to two equivalents of water [44, 46]. When the protein is reduced to the diferrous state, a Glu residue that coordinates one of the irons shifts its position and hydrogen bonds with solvent. A His residue replaces the Glu as an iron ligand. This new geometry of the diiron center enables the reduction of H₂O₂ [40, 41]. This process is reversed with return of the Glu ligand when the irons return to the diferric state [40, 44, 47].

$$\text{NAD}(\text{P})\text{H}+{\text{H}}^{+}+{\text{H}}_2{\text{O}}_2\to \text{NAD}{(\text{P})}^{+}+2{\text{H}}_2\text{O}$$ (1)

4. Redox Enzymes

4.1. Flavodiiron proteins

Flavodiiron proteins (FDPs) are present throughout nature and typically function as NO or O₂ reductases. They are named FDP because they possess an FMN-binding domain and an oxo-bridged diiron site, which resides in a β-lactamase domain. Many FDPs have additional domains with additional functional groups. FDPs are classified as Groups A-H according to the composition of the additional domains [8, 48]. Group A proteins have no additional domain. Group B proteins have a rubredoxin-like domain with a relatively large spacing of amino acid residues between the two Cys-XX-Cys motifs. Group C proteins have an NAD(P)H:flavin oxidoreductase-like (FlvR) domain. Group D proteins have a rubredoxin-like domain with a relatively short amino acid spacing between Cys-XX-Cys motifs. Group E proteins have an iron-sulfur center in addition to three conserved Cys in a CXHX₃C-X₃₂CP motif. Group F proteins have an NADH:rubredoxin oxidoreductase-like domain and a rubredoxin-like domain. Group G proteins have an FlvR and a rubredoxin-like domain. Group H proteins have an FlvR and two rubredoxin-like domains.

To provide an example of this diverse group of FDPs, the FlRd protein from E. coli will be described. This protein is a cytoplasmic O₂ and NO reductase with a higher affinity for NO [49]. FlRd denotes flavo-rubredoxin as the enzyme contains a rubredoxin-like domain in addition to the FMN binding domain, and an oxo-bridged di-iron site. FlRd acts as a NO scavenger under anaerobic conditions and its expression is upregulated under oxidative stress [50]. FlRd is a homotetramesr in solution, as shown by small angle light scattering [51]. The structure of the protein was solved by a combination of x-ray crystallography and NMR [52]. The irons in the oxo-bridged diiron site are coordinated by four His, two Asp and a Glu that acts as a bridging ligand (Figure 6A).

Figure 6.

The FlRd FDP from E. coli A. The oxo-bridged diiron site from PDB entry 5LMC [52]. B. Electron transfer path from NADH-reduced FlRd reductase to the oxo-bridged diiron center for reduction of NO. Two electrons are required for the reduction of NO. They are transferred one at a time via the FMN. Figure 6A was prepared using PyMOL [17].

NADH is the source of electrons for the reduction of NO. Electrons are transferred from NADH to the rubredoxin-like domain by a FAD-dependent FlRd reductase, and then to the diiron center via the FMN (Figure 6B). Electrons are transferred one at a time, given evidence of the formation of the FMN semiquinone without detection the hydroquinone form. The diferrous site then reduces NO as shown in Figure 6B, or alternatively can reduce O₂ [51, 53].

4.2. Ribonucleotide reductase

Ribonucleotide Reductase (RNR) catalyzes the reduction of ribonucleotides to deoxyribonucleotides (Figure 7A). This is required by all organisms for DNA biosynthesis and repair [9, 54]. RNR is also a target for antineoplastic drugs [55] including gemcitabine [56], clofarabine [57], cladribine and fludarabine [58]. RNRs are divided into three classes: I, II and III. The class I RNRs, which are the most studied group of enzymes, are further classified as Ia, Ib and Ic. These three classes are differentiated, in part, by the type of metal center used for catalysis. The class Ia enzymes utilize an oxo-bridged diiron center.

Figure 7. A.

A. Substrate and product of the reaction catalyzed by ribonucleotide reductase. The R group represents the nitrogen base from adenine, cytosine, guanine, or uracil. B. The diiron site in the R2 subunit of the E. coli class ribonucleotide reductase from PDB entry 1MXR [59]. The Fe, N and O atoms are colored brown, blue and red, respectively. Figure7B was prepared using PyMOL [17].

The class I RNRs are α₂β₂ heterodimers. The α subunits, which are named R1, contain the catalytic ribonucleotide reduction site and binding sites for allosteric regulation. The β subunits, which are named R2, house the diiron sites and a critical Tyr residue that functions in a diferric tyrosyl radical complex. This species is the starting point for a long-range electron/radical transfer pathway to oxidize an active-site Cys to initiate catalysis. The oxo-bridged irons are coordinated by two His, three Glu and an Asp (Figure 7B). The diiron site is packed within a typical 4-helix bundle, as well as several additional α-helices (Figure 8), which distinguishes this protein from most other oxo-bridged diiron proteins [59, 60].

Figure 8.

Structure of ribonucleotide reductase. Two orientations are presented to view α₂β₂ heterodimer. The R1 dimer subunits are colored gray and light blue. Only one of the α subunits (light blue) makes contact with the R2 dimer subunits (pale-cyan and pale-green). In this structure, the substrate GDP molecule (red) is present on this α subunit. The two diiron sites on each β subunit are shown as brown spheres. This figure was prepared in PyMOL [17] using PDB entry 6W4X [61].

Structural information of the full α₂β₂ complex has been difficult to obtain by x-ray crystallography because of transient dynamic subunit interactions. This led to uncertainty of the exact mechanism and pathway for the long-range protein coupled electron transfer (PCET) that is required for catalysis. The first holo-complex structure revealed the flexibility of regions involved in this [60]. Recently, the structure of the α₂β₂ complex was obtained by cryo-electron microscopy of the protein (Figure 8) [61].

In the E. coli enzyme, the PCET is initiated in the diiron site with formation of a tyrosyl radical (Fe^III₂-122Y•) in the β subunit. The long range PCET processes then generates a Cys thiol radical (439C-S•) in the α subunit active site that reduces the substrate [61]. In the E. coli enzyme structure, the direct distance between the Tyr and Cys in the structure is approximately 33 Å. The PCET pathway via intervening residues is longer. Several studies have implicated the participation of the residues βTyr122, βTyr356, αTyr731, αTyr720, and αCys439 in the PCET, in that order [55, 61–63]. A role for βTrp48 has also suggested [64, 65]. The hole transfer to the Cys leads to abstraction of a hydrogen from the ribonucleotide, and a formation of a radical intermediate. This leads to production of the deoxynucleotide product and regeneration of the cysteinyl radical [66]. The regeneration of the initial Tyr radical that is associated with the diiron site is achieved by electron/radical transfer in the reverse direction from the Cys radical.

4.3. Methane Monooxygenase

Two types of methane monooxygenases exist. One is a copper-dependent membrane associated particulate methane monooxygenase. The other is a cytoplasmic soluble methane monooxygenase (sMMO) that utilizes an oxo-bridged diiron center. The sMMO complex contains three components: a hydroxylase (MMOH), a FAD-containing reductase (MMOR) and the regulatory protein MMOB. The structure of the M. capsulatus MMOH crystallizes as an α₂β₂γ₂ dimer, as shown in Figure 9A [67]. The diiron center is in the α subunit within a canonical 4-helix bundle. The irons are coordinated by two His residues and four Glu residues (Figure 9B).

Figure 9.

The soluble methane monooxygenase from M. capsulatus Bath. A. The overall structure of the enzyme with its subunits α, β and γ colored in cyan, purple, and gray, respectively. B. The oxo-bridged diiron site of MMOH subunit α. The irons colored brown, oxygen red (including a crystallographic water) and nitrogen blue. Figures were prepared with PyMOL [17] using PDB entry 1MMO [67].

The MMOH activity is stimulated by MMOB and inhibited by an additional regulatory protein, MMOD. Electron transfer in MMOR-MMOH is stimulated by the MMOB-MMOH interaction. Methanol inhibits activity allosterically [68]. The MMOB-MMOH binding induces conformational changes in MMOH that facilitates access of its substrates, CH₄ and O₂, to the active site [69]. The gene that encodes the MMOD inhibitor is located in the sMMO operon. The biding of two MMOD molecules to one MMOH [70] leads to conformational changes in MMOH that disrupt its diiron coordination [71]. The reaction is also affected by the ternary complex formed from MMOH, MMOR, and MMOB, to promote the transfer of two electrons. [69].

MMOH catalyzes the conversion of methane to methanol. Several studies have identified multiple intermediate states of the diiron site that occur during the reaction [10, 72–74]. A mechanism for this reaction is shown in Figure 10 that depicts the structures of some of these intermediates to highlight the changes that occur in the oxo-bridged diiron site during the reaction. MMOR reduces MMOH to form a diferrous center as shown in the first step. The reduction is accompanied by a shift in the coordination of carboxylate of one of the Glu ligands from monodentate to bidentate at the diiron center. Binding of O₂ is facilitated by the interaction between MMOH and MMOB. The state in which O₂ has entered the enzyme prior to interaction with the diiron site is referred to as Intermediate O (not shown). Reaction of O₂ with the diiron leads to the formation of Intermediate P with loss of water (Figure 10). The cleavage of the O-O bond of the bridging peroxo group in the diferric Intermediate P, in concert with the oxidation of both irons to the Fe(IV) state leads to formation Intermediate Q. Intermediate Q is the active species and oxidizes methane. Intermediate R (not shown) is formed from hydrogen abstraction from the methane substrate with one of the irons becoming Fe(III). Then methanol is formed using one of the bridging oxygens with the second iron becoming Fe(III). This intermediate with methanol still present in the enzyme is Intermediate T (not shown). The exact in a manner by which this conversion occurs is not well understood. Addition of water and release of methanol reforms the oxidized MMOH.

Figure 10.

Proposed mechanism of soluble methane monooxygenase.

5. Ferritins

Ferritin (Ftn) is the major protein utilized for iron storage. In doing so, Ftn prevents significant levels of potentially dangerous free iron from accumulating and controls the release of iron when needed. In addition to this important role in iron homeostasis, Ftns and Ftn-like proteins also have roles in detoxification, prevention of oxidative damage, nucleic acid protection, and prevention of nutritional stress [11]. Genetic mutations in the ferritin gene leading to dysfunctional Ftn proteins can cause ferritinopathy and neurodegenerative diseases [75]. Many Ftns self-assemble into 24-subunit oligomers with a hollow inner cage that stores iron atoms (Figure 11A) [76, 77]. This design has been exploited for applications in biotechnology, as a nano-carrier for drug delivery [78] and in biocatalysis [79]. Ftns are present throughout nature in animals and bacteria and share very similar structural features. This section will focus on the human Ftns, and Ftn and Ftn-like proteins from E. coli that exhibit a more diverse range of functions.

Figure 11.

Human H-type Ferritin. A. Structure of the 24-mer structure with two orientations shown to illustrate the 3-fold and 4-fold pore symmetry highlighted in pale-cyan and light blue respectively (PDB entry 6JPS) [80]. B. The 5-helix subunit with A, B and C irons as spheres (PDB entry 4OYN) [82]. C. The oxo-bridged AB diiron site and C iron site. Residues that coordinate the AB site irons are colored in pale-cyan and those that coordinate the C-site iron in light blue (PDB entry 4OYN). [82]. The Fe, N and O atoms are colored brown, blue and red, respectively. Figures were prepared using PyMOL [17].

5.1. Human Ftn

The structure of Ftn monomers consists of a four left-hand twist α-helix bundle with a centralized oxo-bridged diiron center that is coordinated primarily by His and Glu residues. An additional small fifth α-helix is also present near the C terminal (Figure 11B). There are three types of Ftn monomers that are named according to their relative molecular masses; heavy (H), medium (M) and light (L). Human Ftn is composed of 24 of these monomers [80]. Different Ftns share similar 24-subunit shell-like shapes. Iron ions and solvent transit in and out of the molecule by several channels formed at the interfaces between subunits that exhibit 3-fold and 4-fold symmetry (Figure 12A). Ftn may exist as a heteropolymer of H and L subunits or a homopolymer of either H or M subunits. Human L and H subunits share 53% amino acid sequence identity and a similar fold, with variation only in the turn, between the fourth and the fifth helix [81]. Each of the M and H Ftn subunits has one ferroxidase center, which contains two iron binding sites, termed A and B, that are positioned in the middle of the 4-helix bundle (Figure 11B) [82]. The site A iron is coordinated by two Glu residues, a His and two waters. The B site iron is coordinated by two Glu residues and two waters. One of the Glu residues shares bidentate coordination with the irons in and so is a component of both the A and B sites [82] (Figure 12C). A bridging oxygen can be coordinated by the irons in the A and B site. A diiron oxidizing reaction occurs at the AB site to form ferric iron for storage within the Ftn. The diameter of the polymer inner cavity where iron is stored is approximately 80 Å. The L subunit is distinct from the H and M as it does not have a typical ferroxidase center with an AB site [82]. The L Ftn subunit is found in long-term iron storage tissues and exhibits a significantly slower iron oxidation rate at the mineral core [81]. The ratio of H to L monomers in the HL heteropolymers is variable [83–87]. While iron is the physiologically relevant metal, Ftns also have been shown to have affinity for other metals including calcium, cobalt, cooper, magnesium and terbium [88].

Figure 12.

Bacterial Ftn iron centers. A. The AB diiron site of E. coli Ftn (PDB entry 4ZTT) [94]. B. The AB diiron site and C iron site of P. aeruginosa Ftn with the AB site colored in pale-cyan and C site colored in light blue (PDB entry 3R2L) [91]. The Fe, N and O atoms are colored brown, blue and red, respectively. Figures were prepared using PyMOL [17].

Human H-type Ftn has a third iron site termed the C site, which is approximately 6.5 Å from the B site (PDB 4OYN) (Figure 11C). The C site weakly binds ferrous iron and it thought to mediate the entrance and exit of ferric iron from the AB site [89]. Iron in this third site is coordinated by Tyr34, Gln58 and Glu61 [82]. While the ferrous irons in the AB site are preferentially oxidized by O₂, in some situations H₂O₂ can be used oxidant [90]. Studies of variant proteins with mutations that prevented iron from binding to the C suggested that the C site might act as an AB site modulator that assists in a H₂O₂ detoxification mechanism [90]. In the human Ftn L subunit, the A and B iron sites are replaced by a mineral core nucleation site, which is pointed to the inner cage surface. This is a trinuclear iron site that is coordinated by Glu residues [81].

5.5. Escherichia coli ferritins

E. coli expresses two Ftns and two Ftn-like proteins. EcFtnA and EcFtnB are archetypal ferritins. EcDps is a DNA-binding protein from cells experiencing starvation. The other is bacteroferritin (EcBfr) [12], which has a b-type heme present in its dimer interface. EcFtnA has only 22% sequence identity to human H-type Ftn, but does share the same overall structure with a 4-helix bundle monomer that assembles into a 24-subunit protein with a hollow shell used for iron storage. It is noteworthy that most of the identical residues are clustered on the center of the 4-helix bundle. The E. coli FtnA also has the AB and C iron sites [90]. In its crystal structure, the iron is observed only in the AB site, which is shown in Figure 12A. However, the three irons in the AB and C site are observed in the crystal structure of the bacterial Ftn from P. aerugiosa (Figure 12B)[91]. EcFtnA lacks a conserved sequence of four Glu residues from the mammalian L subunit that has been associated with the iron core nucleation, however it retains the iron nucleation and its function as an iron storage protein [92]. EcFtnB has low sequence similarity to with FtnA and lacks residues important for fast iron oxidation [93]. Its shares only 25% of primary sequence identity with human Ftn L, and the core mineralization residues of the L-type ferritins are absent in EcFtnB.

EcDps binds DNA and protects it from oxidative stress by neutralizing H₂O₂. Analogous proteins are found in eubacteria and archaea [95]. It is composed of 12, rather than 24 subunits and is able to store up to 500 irons per molecule shell [96]. In contrast to other Ftns that use O₂ as the major electron acceptor, EcDps prefers H₂O₂ as an oxidant for the ferroxidation and mineralization reactions [97]. In vivo studies showed that a Trp52Ala mutation decreased bacterial resistance to the potentially toxic H₂O₂ [98]. The lysine enriched N-terminal of EcDps binds to DNA. This allow EcDps to protect DNA from oxidative damage [99, 100]. M. tuberculosis also has a DPS Ftn-like protein that has this activity [101].

The EcBfr has 24 identical H subunits and up to 12 protoporphyrin IX groups, one per dimer interface. A conserved Met binds the b-type heme to the polypeptide (Figure 13A). In addition to the 3- and 4-fold symmetry channels, EcBfr also has another channel named the B-pore. The EcBfr has AB and C sites similar to those described in other ferritins (Figure 13B). Bfrs, which are composed of heterodimers, are also found in Pseudomonas aeruginosa [102] and M. tuberculosis [101]. The BfrA from M. tuberculosis is unique as it has both Dps-like function and catalase activity. The catalase function might counteract the absence of the Dps gene in this bacterium, thus preventing H₂O₂ oxidative damage [101].

Figure 13.

Bacterioferritin from E. coli. A. Overall structure of the protein with the subunits of the dimer colored light blue and gray. The irons are shown as spheres and the heme is colored in red. B. Residues that coordinate the AB diiron and the C site. Figures were prepared with PyMOL [17] using PDB entry 3E1M [103].

6. Summary

The oxo-bridged iron proteins described in this review share a common structural fold, a 4 α-helix bundle that houses the oxo-bridged diiron site. However, the overall structures of the different proteins that share this feature can be quite different, and the identity of the ligands that coordinate the irons varies. These proteins exhibit a wide range of functions including iron storage, oxygen transport, sensing and a wide range of catalytic activities employing different reaction mechanisms. Recognition of these types of proteins is growing. This is reflected by the recent increase in identification of a variety of hemerythrin-like proteins with diverse activities, including a distinct subset in mycobacteria. While the structural characterization of these proteins has significantly advanced, much still remains to be determined regarding the precise physiological roles and mechanisms of action of many of the more recently identified proteins.

Abbreviations:

Bfr

bacterioferritin

HLP

hemerythrin-like protein

FDP

flavodiiron protein

Ftn

ferritin

MMO

methane monooxygenase

PCET

protein coupled electron transfer

RNR

ribonucleotide reductase