The Iron Tug-of-War between Bacterial Siderophores and Innate Immunity

Abstract

Iron is necessary for the survival of almost all aerobic organisms. In the mammalian host, iron is a required cofactor for the assembly of functional iron-sulfur (Fe-S) cluster proteins, heme-binding proteins and ribonucleotide reductases that regulate various functions, including heme synthesis, oxygen transport and DNA synthesis. However, the bioavailability of iron is low due to its insolubility under aerobic conditions. Moreover, the host coordinates a nutritional immune response to restrict the accessibility of iron against potential pathogens. To counter nutritional immunity, most commensal and pathogenic bacteria synthesize and secrete small iron chelators termed siderophores. Siderophores have potent affinity for iron, which allows them to seize the essential metal from the host iron-binding proteins. To safeguard against iron thievery, the host relies upon the innate immune protein, lipocalin 2 (Lcn2), which could sequester catecholate-type siderophores and thus impede bacterial growth. However, certain bacteria are capable of outmaneuvering the host by either producing “stealth” siderophores or by expressing competitive antagonists that bind Lcn2 in lieu of siderophores. In this review, we summarize the mechanisms underlying the complex iron tug-of-war between host and bacteria with an emphasis on how host innate immunity responds to siderophores.

Introduction

Iron is the second most abundant element in the Earth's crust and the most bountiful metal ion in the human body [1, 2, 3]. This indispensable nutrient is necessary for the survival of almost all aerobic organisms, excluding Lactobacilli [4] and Borrelia burgdorferi [5]. The majority of iron in mammals is recycled from the hemoglobin liberated from senescent red blood cells, with only a small portion obtained from the diet [1, 6]. When complexed to heme, iron-sulfur (Fe-S) clusters and ribonucleotide reductases, both ferrous (Fe²⁺) and ferric (Fe³⁺) iron partake in multitudinous biological functions, including oxygen transport, cellular respiration and DNA synthesis [6, 7]. Moreover, this electron carrier participates as a redox catalyst in the Fenton and Haber-Weiss reactions, generating reactive oxidative species (ROS) [6, 7]. Despite its multifaceted roles, iron can be harmful at high levels due to its toxicity and its capacity to generate oxidative stress; therefore, its bioavailability in the mammalian host is stringently regulated throughout its absorption, transport and storage. Under aerobic conditions, iron accessibility is naturally limited since the predominant, soluble ferrous iron is oxidized to its insoluble ferric form, which is further polymerized to ferric (oxy)hydroxides [1, 8]. Moreover, the majority of iron present in circulation is tightly bound to transferrin, leaving only a small fraction of free iron (approx. 10^–24 M) [1], which is insufficient to support bacterial growth that requires 10⁵–10⁶ M iron per cell [9]. During inflammation and/or infection, the host could further deplete circulating iron by inducing an acute phase response and upregulating various inducible mechanisms to withhold essential, mineral nutrients against potential pathogens [10, 11]. Such strategies that limit nutrient accessibility to microbes are collectively termed as “nutritional immunity.”

To countervail nutritional immunity, there are two main strategies that commensal and pathogenic microbes employ for iron acquisition. One method is for bacteria to directly acquire iron from host proteins, such as transferrin and heme, in a receptor-mediated manner [12]. Bacteria utilize their outer membrane receptors that can bind and chelate iron directly from transferrin; likewise, bacteria can synthesize hemophore proteins that are able to bind heme and transport the host protein to the microbial environment through receptor uptake [12]. Analogous to hemophores, Gram-negative (e.g., Escherichia coli, Pseudomonas aeruginosa, Klebsiella pneumoniae), Gram-positive (e.g., Staphylococcus aureus), and Mycobacteria (e.g., Mycobacterium tuberculosis) [13] apply the second method of iron acquisition by synthesizing and secreting diverse siderophores. Siderophores are low molecular weight (500–1,500 Da) iron chelators that have a potent affinity for iron compared to the host iron-binding proteins [3]; moreover, these chelators are selective for the insoluble ferric iron but not the soluble ferrous iron [3]. Taking into account the positively charged iron, siderophores utilize their deprotonated, negatively charged oxygen molecules for the strongest interaction [3]. Most siderophores, with the exception of vibriobactin (Vbb) [14, 15], use six donor oxygen atoms that form an octahedral structure to encapsulate the iron, producing a hexadentate center [3]. This stable tri- or tetradentate complex permits interaction with the siderophore-specific receptor, followed by engulfment and hydrolysis by an esterase (e.g., Fes, IroD, IroE) to release the apo-siderophore and ferric iron (siderophore structures and transport have been well reviewed in previous articles [1, 3]). In this review, iron-bound siderophores will be referred to as either iron-laden or holo-siderophores, whereas iron-free siderophores will be mentioned as apo-siderophores for convenience.

Since the initial discovery of the first three siderophores, mycobactin, ferrichrome and coprogen, throughout 1949–1952, more than 500 distinct siderophores have been identified [3]. Aside from the siderophores characterized in bacteria, fungi and plants, recent studies have also identified mammalian siderophore-like biomolecules, such as norepinephrine [16] and 2,5-dihydroxybenzoic acid (2,5-DHBA) [17] that also partake in iron transport and homeostasis. Currently, there are five structural classifications for the iron chelators: catecholate, hydroxamate, carboxylate, phenolate and mixed siderophores [3]. The structural differences among siderophores portray the virulence of microbes, especially in regards to whether their siderophores could be sequestered, and thus subdued, by the host innate immune protein, lipocalin 2 (Lcn2; alias siderocalin/SCN, human ortholog neutrophil gelatinase-associated lipocalin/NGAL, mouse ortholog 24p3, uterocalin). To resist Lcn2-mediated inhibition, pathogens such as Salmonella spp. and Klebsiella spp. have evolved to express stealth siderophores (e.g., salmochelin, aerobactin, and yersiniabactin), which augment siderophore-mediated iron acquisition and, therefore, microbial virulence and pathogenicity [3]. This thrust and parry between siderophores and the host innate immune response, respectively, and their therapeutic potentials will be further explored in this review.

Nutritional Immunity: Hide and Seek Battle between Host and Microbes

During infection and/or inflammation, the host employs a nutritional immune response to prevent bacterial growth and to limit iron from causing unwarranted oxidative stress and injury to the host [6]. At the same time, however, diminished iron availability can decrease the efficacy of iron-dependent antimicrobial defense systems for innate immune cells during infection [6]. This section of the review will discuss the inducible mechanisms, including hepcidin upregulation and ferric iron-binding proteins, and their physiological impacts due to attenuated free iron levels.

Hepcidin-Induced Degradation of Ferroportin Promotes Hypoferremia

There are multiple mechanisms that the host employs to hinder iron accessibility when a potential pathogen is detected. At the onset of infection and/or inflammation, the host upregulates hepatic interleukin (IL)-6 which, in turn, induces the expression of hepcidin [18, 19]. Hepcidin is a peptide hormone known for its role as the master regulator of iron homeostasis [19]. Upon release into systemic circulation, hepcidin binds to ferroportin (Fpn) expressed on macrophages and the duodenal epithelia, which consequently induces the internalization and degradation of Fpn [19]. Fpn is the sole exporter of nonheme iron in mammals; hence, its downregulation by hepcidin traps iron within macrophages and, moreover, prevents iron absorption from the intestines [20]. This course of events is regarded as the primary mechanism of initiating hypoferremia or anemia of inflammation [18, 19], which can result in the iron starvation of extracellular pathogens and thus inhibit bacterial growth. Incidentally, the innate protein, Lcn2, was found to also participate in stimulating hypoferremia [21, 22], though further studies are needed to elucidate whether this occurs complementarily or independently of hepcidin.

Yet, for intracellular pathogens such as Salmonella spp. that primarily infect macrophages, the hepcidin-driven hypoferremic response appears to be largely beneficial. Accumulating evidence has shown that S. enterica serovar Typhimurium infection can prompt the hepatic secretion of hepcidin through activation of IL-6 signaling [23]. The increased retention of iron in macrophages provides an iron-rich haven for Salmonella to augment their growth. To thwart Salmonella from hijacking their cellular iron, macrophages counteract by upregulating Lcn2 and Fpn in response to interferon-γ following infection [24]. Despite the advantages of the nutritional immune molecules, it is interesting to note that, in the context of Streptococcus pneumoniae intracellular infection, high levels of Lcn2 can also be detrimental by impeding the macrophage immune response and pathogen clearance [25].

Ferric Iron-Binding Proteins Supply the Fuel for ROS Generation

In addition to regulating the expression of iron transporters, the host utilizes both basal and induced ferric iron-binding proteins to sequester iron availability. As mentioned previously, at physiological conditions most of the host-free iron is bound to transferrin, which can then transport iron to cells via endocytosis of the transferrin receptor (TfR1). The transfer of iron from transferrin to cytoplasmic storages (e.g., ferritin) promotes hypoferremia and restricts iron from extracellular bacteria. Moreover, the delivery of iron to innate immune cells, such as neutrophils and macrophages, is crucial as these cells require iron to fuel their oxidative responses during intracellular infection. To further efficiently deplete free iron, neutrophils release the ferric iron-binding protein, lactoferrin, which has a greater affinity for iron than transferrin [26]. Iron-laden lactoferrin can be taken up by both immune and nonimmune cells via a receptor-mediated, clathrin-dependent endocytosis that is analogous to transferrin uptake [27]. The delivery of iron to infected cells via lactoferrin and transferrin supplies the catalyst for the Haber-Weiss and Fenton reactions, where ROS generation enhances bactericidal activity through oxidative killing. However, the affinity of lactoferrin and transferrin for iron is inferior compared to siderophores, thus putting the formers at a disadvantage against siderophore-producing bacteria. By wrestling iron from host proteins, siderophores not only acquire a vital nutrient for bacteria, but also impede host ROS generation and other antimicrobial responses that are iron dependent.

Bacterial Siderophores: Iron Thievery Weapons

Unlike the mammalian host which utilizes iron as a catalyst to produce oxidative antimicrobial responses, bacteria require iron for redox-sensitive transcription factors and Fe-S-containing regulatory proteins that work to detoxify ROS like nitric oxide (NO) [28, 29, 30, 31, 32, 33]. In addition to regulating NO metabolism, iron levels control the bacterial transcription repressors in the ferric uptake regulator protein (Fur) family [34, 35]. Under iron excess conditions, ferrous iron is complexed to Fur, which allows Fur to repress the genes for siderophore biosynthesis [34, 35]. During nutritional immunity, deficient iron levels inhibit Fur binding to siderophore genes, thus allowing for transcriptional activation of siderophore synthesis [34, 35]. The following section will discuss the three more prevalent siderophore-types and their virulence through Lcn2-mediated interaction (summarized in Table 1).

Table 1.

Siderophore virulence through Lcn2-mediated interaction

Classification	Example(s)	Bacterial origin(s)	Lcn2 resistant (yes, no, undefined)	Reference(s)
Catecholate	Ent	E. coli	No	3, 13, 37, 38
		K. pneumoniae P. aeruginosa
	Bcb	B. anthracis	No	3, 37
		B. subtilis
	Ptb	B. anthracis	Yes	3, 37
	Vbb	V. cholerae	Undefined	3, 14, 15
Trihydroxamate	DFO	S. pilosus	Yes	3, 39, 40
	Exochelin	M. smegmatis	Yes	3, 41, 42
Mixed	Mycobactin	M. smegmatis	Yes	3, 41, 42
	Carboxymycobactin	M. smegmatis	No	3, 41, 42
		M. tuberculosis
	Aerobactin	E. coli	Yes	3, 13
		K. pneumoniae
	Yersiniabactin	K. pneumoniae	Yes	3, 13
	Pyoverdine	P. aeruginosa	Yes	3, 13
	Pyochelin	P. aeruginosa	Yes	3, 13
	Salmochelin	E. coli	Yes	3, 45
		K. pneumoniae

The table depicts the three prevalent categories of siderophores, including catecholate, trihydroxamate, and mixed siderophores. For each iron chelator category, a few examples are provided including information regarding their bacterial synthesis origin and whether they are lipocalin 2 (Lcn2) resistant. In general, catecholate-type siderophores are inhibited by Lcn2, except for petrobactin. Trihydroxamate and mixed siderophores are Lcn2 resistant, with the exception of carboxymycobactin. Ent, enterobactin; Bcb, bacillibactin; Ptb, petrobactin; Vbb, vibriobactin; DFO, deferoxamine.

Catecholate Siderophores

Well-known catecholate siderophores include the prototypical enterobactin (Ent; alias enterochelin), bacillibactin (Bcb), petrobactin (Ptb) and Vbb. Their structures, illustrated in Figure 1a, depict analogous catechol-chelating subunits. Except for Ptb possessing two 3,4-dihydroxybenzoate-chelating sites, the others share the common three 2,3-DHBA-chelating moieties (Fig. 1d) [3]. Compared to other categories of siderophores, the catecholate-type has stronger iron affinities than transferrin and lactoferrin, where Ent has the most potent affinity with a K_D of 10⁻³⁵ M at physiological pH [3]. In response to hypoferremia and inflammation, the host secretes the acute phase protein, Lcn2, to chelate specifically catecholate and some carboxylate siderophores [36]. Through tryptophan fluorescence quenching, it has been shown that Ent and Bcb interact with Lcn2 in an equimolar stoichiometry [37, 38] and, therefore, have low virulence. Even though Ptb is a catecholate siderophore, its 3,4-dihydroxybenzoate-chelating subunit renders Lcn2 unable to bind, thus amplifying its virulence [3]. Vbb has the same 2,3-DHBA-chelating subunits as Ent and Bcb, but instead of harnessing six oxygen atoms for iron encapsulation, this siderophore utilizes one nitrogen and five oxygen atoms to form the six ligands that envelop the ferric iron [14]. Whether Lcn2 can sequester Vbb needs further investigation as one study proposes that they interact with low affinity [14], while another reports a strong association between the two [15].

Fig. 1.

Siderophore structures. a The bacterial catecholate siderophore structures of Ent, Bcb, Ptb and Vbb. b The structures for the bacterial hydroxamate siderophores, DFO and exochelin. c Diagrams of the mixed siderophores, including aerobactin (citryl-hydroxamate), yersiniabactin (phenolate/thiazoline), pyoverdine (catecholate-hydroxamate), pyochelin (phenol-catecholate), mycobactin (hydroxyphenyloxazoline derived from salicylic acid), and carboxymycobactin (carboxylate of mycobactin). d The proposed mammalian siderophores 2,5-DHBA and norepinephrine. These are suggested siderophores due to their structural similarities with 2,3-DHBA of Ent.

Trihydroxamate Siderophores

As Lcn2 is, more or less, catecholate specific, the immune protein is unable to sequester hydroxamate and mixed siderophores. A well-known trihydroxamate is deferoxamine (DFO; desferrioxamine; Fig. 1b), a nonpathogenic, bacterial siderophore produced by Streptomyces pilosus [39]. This hexadentate chelator binds iron in a 1: 1 ratio, where its hydrophilic nature renders it difficult to penetrate cell membranes (except hepatocytes) and to be sequestered by Lcn2; thus, it is mostly secreted in urine and feces [40]. Due to the iron excretion capabilities of DFO, this siderophore has been utilized as an agent (clinically known as Desferal) for iron-chelation therapy, primarily in the treatment of patients with iron-overload disorders like hemochromatosis and β-thalassemia [40]. Exochelin is another extracellular trihydroxamate pentapeptide derivative (Fig. 1b) secreted by saprophytic, nonpathogenic mycobacteria (e.g., M. smegmatis) [41]. Aside from the nonextractable, water-soluble exochelin, M. smegmatis secretes another chloroform-extractable siderophore that shares more similarity to its intracellular mixed siderophore, mycobactin (Fig. 1c) [41]. However, instead of a long alkyl chain, as shown with mycobactin, the chloroform variant contains a short acyl chain that ends with a carboxylic acid group [41]. Due to this structural distinction, the name carboxymycobactin (Fig. 1c) was given to distinguish the two mixed siderophores. Carboxymycobactin was also found to be secreted by various pathogenic mycobacteria (e.g., M. tuberculosis) indicating high virulence [41]. Yet, carboxymycobactin is among the few, if not the only, mixed siderophore identified to date that can be sequestered and thus inhibited by Lcn2 [42].

Mixed Stealth Siderophores and Xenosiderophores

In contrast to carboxymycobactin, diverse pathogens have evolved to produce the so-called stealth siderophores to evade Lcn2 inhibition, which include prevalent mixed siderophores such as aerobactin, yersiniabactin, pyoverdine and pyochelin depicted in Figure 1c [3]. Due to being Lcn2 resistant, aerobactin has been shown to enhance the hypervirulent phenotype of K. pneumoniae in lung infections [43]; likewise, yersiniabactin promotes the virulence of the bubonic plague pathogen, Y. pestis [44]. In addition to these naturally synthesized mixed siderophores, certain bacteria can modify an existing Lcn2-sensitive siderophore into a stealth siderophore as another evasive strategy. For instance, Salmonella spp., Klebsiella spp., and certain strains of pathogenic E. coli can modify Ent into salmochelin by adding two additional glucosyl residues. Unlike Ent, salmochelin is too bulky to fit in the hydrophobic pocket of Lcn2 due to steric hindrance [45]. Aside from relying on stealth siderophores, certain bacteria instead express the receptors for nonnative siderophores expressed by other bacteria, thus allowing the former to exploit “xenosiderophores” expressed by the latter. For example, P. aeruginosa has multiple siderophore-uptake systems capable of utilizing aerobactin, despite lacking the genes needed for aerobactin synthesis itself [46].

Mammalian Siderophore-Like Biomolecules: A Renegade in the Tug-of-War for Iron?

In addition to bacteria, fungal (e.g., ferrichrome) and plant (e.g., EGCG) siderophores [47], it has been recently proposed that mammals can also synthesize and secrete endogenous biomolecules that are analogous to siderophores. Similar to their bacterial-derived counterparts, mammalian siderophore-like biomolecules also participate in the host-bacteria tug-of-war for iron. However, there has been contrasting results as to whether these biomolecules are beneficial or detrimental to the host. This section of the review will explore the two proposed types of mammalian “siderophores” and how they impact host iron homeostasis.

2,5-Dihydroxbenzoic Acid

One of the proposed mammalian biomolecules is 2,5-DHBA (gentisic acid) due to it having an iron-binding moiety that is similar to the 2,3-DHBA monomeric subunit of Ent (Fig. 1d) [17]. Additionally, the enzyme, butyrate dehydrogenase-2 (BDH2) that synthesizes 2,5-DHBA in mice, is reported to be a homolog of the EntA protein that catalyzes 2,3-DHBA in bacteria [17]. Considering the strong similarities between 2,5-DHBA and Ent, studies have explored as to whether the biomolecule can interact with Lcn2. Some studies that used tryptophan fluorescence quenching have reported strong binding between holo-2,5-DHBA and Lcn2 [17, 48], whereas others did not see any difference in quenching indicating weak binding [49]. Irrespective of Lcn2 interaction, it has been shown that 2,5-DHBA could augment bacterial growth under Ent-depleted conditions [48], suggesting that pathogens could utilize host-derived 2,5-DHBA as a source of iron. To prevent this thievery, bdh2 expression is diminished after TLR agonists, especially lipopolysaccharide, activate TLR-mediated pro-inflammatory signaling on macrophages [48].

On the opposite spectrum, studies have shown that the loss of BDH2 in mice reduces iron levels in the mitochondria, proposing that 2,5-DHBA has a role in iron transport to decrease ROS and utilize the iron for heme synthesis, electron transport and other functions [17, 50]. However, a recent report has shown that 2,3-DHBA, Ent and 2,5-DHBA can augment mitochondrial ROS generation and attenuate oxidative phosphorylation in cardiomyocytes, which was further potentiated by apo-Lcn2 [51]. To better understand the role of 2,5-DHBA in iron homeostasis and mitochondrial bioenergetics, future studies need to confirm if Lcn2 can complex with holo-2,5-DHBA. If so, the next step would be to determine if this complex can bind to lipocalin receptors (e.g., 24p3R, megalin) and be transported to the mitochondria. Likewise, upcoming experiments should determine whether enhanced Lcn2 expression can inhibit 2,5-DHBA-mediated bacterial growth under siderophore-deficient conditions. If holo-2,5-DHBA does not complex with Lcn2, then research is warranted to determine if there is an Lcn2-independent transportation mechanism for 2,5-DHBA to enter the mitochondria.

Catecholamine

Aside from 2,5-DHBA, host-derived catecholamine (e.g., norepinephrine; Fig. 1d) also possesses a catechol moiety that is similar to Ent [1]. Norepinephrine has been shown to abstract iron from transferrin and lactoferrin, and thereafter could either shuttle the iron to siderophores or be directly taken up by bacteria [16]. Interestingly, in parallel to catecholate-type siderophores, Lcn2 can sequester holo-L-norepinephrine, and thus prevent bacteria from exploiting host catecholamine as a source of iron [16]; likewise, catechol itself has been shown to chelate and deliver iron to cells via Lcn2 interaction [52]. Considering that catechol can be derived from both mammalian and bacterial metabolisms, this represents a host/microbiome interaction [52]. Overall, further studies are certainly needed in order to fully understand the role of these putative endogenous mammalian biomolecules in host physiology, as well as their impact on the host-bacteria tug-of-war for iron.

Innate Immunity Counters Siderophore Iron Thievery

When a pathogen is uncovered in the mammalian host, the first line of defense is the innate immune system. This comprises the recruitment of innate immune cells (e.g., neutrophils, macrophages) that secrete antimicrobial proteins and enzymes to assist in phagocytosis and killing of the pathogen. The following section will discuss how innate immune proteins directly inhibit siderophore-mediated iron-chelation, which is further complemented by expressions of cytokines and chemokines.

Lipocalins Sequester Catecholate-Type Siderophores

Lipocalins are a diverse group of proteins sharing a highly conserved core structure that can bind small, hydrophobic ligands [53]. Their core structure consists of an eight-stranded anti-parallel β-barrel that defines the three calyx/cup binding pockets [53]. The human 25-kDa Lcn2 was first identified in 1993 as a protein that formed a complex with the matrix metalloproteinase 9 [54, 55]. In 2002, Goetz et al. [56] discovered that Lcn2 could bind tightly to the iron-laden, catecholate-siderophore, Ent, via a combination of simple ionic and cation-π interactions. It is noteworthy to emphasize that Lcn2 itself does not bind to iron directly, thus providing it with the advantage of not having to compete with the unmatched affinity of Ent towards iron. In 2004, Aderem and colleagues [38] unraveled that Lcn2 is an innate immune protein and a bacteriostatic agent with a specialized role in sequestering siderophores to prevent bacterial iron acquisition (Fig. 2).

Fig. 2.

Iron tug-of-war: how Lcn2 and MPO interact with siderophores. MPO is secreted from the primary granules (1⁰) of neutrophils and utilizes hydrogen peroxide and halide ions to form hypochlorous acid (HOCl) that serves as an oxidizing and antimicrobial agent. Apo-Ent inhibits full activation of MPO, thus blunting HOCl production; moreover, apo-Ent also prevents the degranulation of MPO and formation of NETs. A second mechanism to thwart siderophores is by secreting Lcn2 from the secondary granules (2⁰) of neutrophils, where Lcn2 binds to holo- or apo-Ent in a 1: 1 ratio. However, some pathogens have evolved stealth siderophores that are insensitive to Lcn2 and thus can supply iron to pathogens.

Under physiological conditions, Lcn2 is secreted constitutively throughout the body, e.g., liver and gut epithelia are among the major sources, in addition to being stored within the secondary granules of neutrophils (Fig. 2) [36]. During inflammation, Lcn2 levels are upregulated by several log orders of magnitude to mediate its functions as an acute phase protein, a neutrophil chemotactic agent and an enforcer of host nutritional immunity (as reviewed by Xiao et al. [36]). Incidentally, we have demonstrated the utility of measuring fecal Lcn2 as a noninvasive, reliable and sensitive marker for gut inflammation [57]. Upregulation of Lcn2 is potentiated through MyD88-dependent NF-κB signaling following the activation of activated pattern recognition receptors (PRR), i.e., TLR4 and TLR5 [38, 58] that recognize bacterial lipopolysaccharide and flagellin, respectively. Accordingly, deficiency in urinary Lcn2 has been associated with increased predisposition to urinary tract infections in humans [59]. Likewise, mice deficient in Lcn2 and humans with neutrophil-specific granular deficiency – a rare disorder characterized by the absence of secondary granules and thus lacking Lcn2 and lactoferrin – are highly susceptible to bacterial infections [60].

Notwithstanding the antimicrobial role of Lcn2, this bactericidal agent is rather limited in its capacity to sequester mostly catecholate and some carboxylate siderophores. Lcn2 can bind either apo- and holo-siderophores in an equimolar stoichiometry ratio (Fig. 2), but with less affinity when binding to apo-siderophores (e.g., at physiological pH, Kd of 0.41 nM for holo-Ent; Kd of 3.57 nM for apo-Ent) [56, 61]. To broaden the range of siderophore inhibition, epithelial cells also secrete tear lipocalin (alias Lcn1) predominantly into tears and respiratory secretions in response to infection [62]. Lcn1 can bind to diverse hydrophobic compounds including fatty acids, phospholipids, cholesterol and siderophores, but do not bind iron directly as analogous to Lcn2 [62]. Lcn1 has a broader range of siderophore interaction than Lcn2, as the former is capable in binding to catecholate and hydroxamate siderophores, except pyoverdine due to its additional bulky chromophore [62]. As hydroxamate siderophores are mostly from fungi, it has been proposed that Lcn1 may be more specialized in eliminating fungal infections [62]. Despite this, this lipocalin has a weaker affinity for siderophores compared to Lcn2 [62], suggesting that Lcn1 may be a reinforcement for Lcn2. The disparity and complementary features of both lipocalins, especially in regard to where they are synthesized and secreted during infection, warrant further investigation.

Siderophore-Mediated Iron-Chelation Promotes an Inflammatory Response

In addition to lipocalins, the host also employs several other strategies to safeguard its iron supply, such as one that involves directly sensing the presence of the siderophore followed by a timely immune response. The catecholate Ent, by itself, has been shown to induce upregulation of IL-8 from cultured human respiratory epithelial cells; such cytokine response was further augmented in the presence of Lcn2 [63]. This pro-inflammatory feature of Ent is intimately tied to its iron-chelating activity as only iron-free Ent, but not iron saturated-Ent, could induce an IL-8 response. The hydroxamate, DFO, can also induce an IL-8 response from epithelial cells in vitro through p38 MAPK (mitogen-activated protein kinase) signaling [64]. Further studies suggest that the ability to trigger an immune response may be a common feature for most siderophores, resulting in the upregulation of predominantly cytokines and chemokines (e.g., IL-6, IL-8, CCL20, Lcn2) to recruit immune cells [65]. Intriguingly, inflammatory responses induced by stealth siderophores, i.e., yersiniabactin and salmochelin, could be further augmented by Lcn2 [65], despite the inability of the former to interact with the latter. Such observations could be explained by the fact that Lcn2 itself is able potentiate a pro-inflammatory response, independently of its role in siderophore binding [37].

In addition, siderophores have been shown to also stimulate transcription factors that are normally inhibited by enzymes that require iron as a cofactor. One such iron-regulated transcription factor, hypoxia-inducible factor 1α (HIF-1α), is only active when HIF-1α in the cytosol translocates into the nucleus and heterodimerizes with HIF-1β [66]. HIF-1α is opposed by prolyl hydroxylase (PHD), which requires either oxygen or iron as a cofactor [66]. PHD hydroxylates the prolyl residues of HIF-1α, which become binding sites for the von Hippel-Lindau tumor suppressor protein, tagging HIF-1α for proteasomal degradation [66]; therefore, HIF-1α is kept inactive under aerobic or iron-saturated conditions. However, siderophores can stabilize HIF-1α through indirect inhibition of PHD via iron starvation, thus allowing activation of HIF-1α in a hypoxia-independent manner [66]. One of the genes regulated by HIF-1α is iNOS (NO synthase), which synthesizes NO from arginine [67]. The production of NO results in two main effects: (i) NO aids the phagocytic activity of macrophages and neutrophils, and (ii) NO inhibits PHD, allowing for continuous stabilization of HIF-1α [66, 67]. Hence, through HIF-1α, siderophores inadvertently augment the bactericidal activity of the innate immune response.

Siderophores Outsmart the Innate Immune Response

The secretion of lipocalins are key in protecting the host from siderophore-mediated iron chelation; however, bacteria have further enhanced their countermoves for iron sequestering, including the generation of stealth siderophores (Fig. 2). The following section discusses how bacteria utilize siderophore competitive antagonists to evade Lcn2-siderophore interaction and how siderophores are exploited to dampen host iron-dependent immune responses.

Cyclic Diguanylate Monophosphate Antagonizes Lcn2 in Lieu of Siderophores

Microbes employ various strategies to evade Lcn2. One method is for certain bacteria, including K. pneumoniae, to downregulate the exportation of Ent in order to upregulate the release of the linear Ent-cleaved product 2,3-dihydroxybenzoylserine (DHBS) [68]. Even though the iron affinity for DHBS is weaker than Ent, this breakdown product can act as an Lcn2-resistant stealth siderophore [68]. In addition to DHBS and other stealth sid erophores, bacteria can upregulate the secretion of the highly conserved second messenger, cyclic diguanylate monophosphate (c-di-GMP), to enhance virulence [69, 70]. C-di-GMP promotes biofilm formation, motility and drug resistance for multiple bacterial species [70]; likewise, loss of c-di-GMP in a S. enteritidis mutant substantially reduced its virulence [71]. Moreover, this second messenger has been known to cross-talk with only two host innate immune proteins, STING and DDX41, where receptor activation leads to stimulation of tank-binding kinase 1 (TBK1). TBK1 then phosphorylates IRF3 (interferon regulatory transcription factor 3), which allows for the generation of type 1 interferons and thus infection elimination [70]. However, in a recent study, Li et al. [70] discovered that c-di-GMP can also interact with Lcn2. Specifically, c-di-GMP interfaces with the same hydrophobic binding pocket that catecholate siderophores also bind on Lcn2; hence, binding of c-di-GMP to Lcn2 renders the latter incapable of sequestering siderophores and thus preserving bacterial iron uptake and growth [70]. Future studies are warranted to explore whether this c-di-GMP/Lcn2 interaction presents a novel PAMP/PRR (pathogen-associated molecular patterns/pattern recognition receptor) sensing mechanism; moreover, future studies are needed to determine how the second messenger interacts with Lcn2, highly contemplating what allows c-di-GMP, but not c-di-AMP, to interface with Lcn2 [70].

Ent Hijacks Neutrophil Bactericidal Mechanisms

Aside from scavenging iron, siderophores are also capable of interfering with host immune responses, particularly those which are iron dependent. One example is the ability of Ent to impede the bactericidal activity of myeloperoxidase (MPO) exclusively expressed in the primary granules of neutrophils (Fig. 2) [72, 73]. MPO is a ferric-dependent heme peroxidase that participates in the antimicrobial activity of neutrophils by utilizing hydrogen peroxide and halide ions, during respiratory burst, to generate hypochlorous acid (Fig. 2) [73]. Specifically, first ferric-MPO is converted to oxoiron (Fe[IV] = O), also known as compound I, when introduced to hydrogen peroxide; second, compound I is reduced back to its ferric state when halide ions are present, and third, HOCl is then produced as the oxidizing and antimicrobial agent to eliminate pathogens (Fig. 2) [73]. By employing bovine lactoperoxidase – which is similar to MPO in function – as a model, Singh et al. [73] performed spectral analysis to examine the effects of Ent on the peroxidase catalyzed reaction with hydrogen peroxide. Results indicated that apo-Ent – and also monomeric 2,3-DHBA to a lesser extent – inhibited the LPO-catalyzed redox reaction of compound I (Fe[IV] = O) conversion to the ferric state [73]. This would in effect blunt the production of antimicrobial reactive intermediates, such as HOCl generated by MPO (Fig. 2). This blockade was specific to apo-Ent, but not iron-laden Ent; moreover, other siderophores including salmochelin, yersiniabactin and ferrichrome did not exhibit the analogous inhibition, even when at higher concentrations [73]. Following this observation, another study by Coker et al. [74] employed a more comprehensive array of MPO activity assays to elucidate the functional interactions between Ent and MPO. The study, however, asserts that Ent is unlikely to inhibit the activity of MPO, at least not directly per se, but instead function as a substrate for MPO-catalyzed reactions [74]. Although Coker et al. [74] did not examine whether Ent could inhibit MPO- and neutrophil-mediated bacterial killing, other studies which document that Ent could do so [72, 73] argue that this siderophore may have other indirect mechanisms to alter MPO and neutrophil activity. Such mechanisms could, in part, be driven by the potent antioxidant and ROS-neutralizing property of Ent [75, 76], which warrant further studies.

Saha et al. [72] further demonstrated that Ent also inhibits ROS generation in both mouse and human neutrophils upon stimulation with PMA (phorbol 12-myristate 13-acetate) in vitro. Additionally, Ent could likewise inhibit the degranulation of primary granules that contain MPO (Fig. 2), but not the secondary granules that contain Lcn2 and lactoferrin [72]. The underlying mechanisms of inhibition are yet to be well understood, though it has been proposed to be mediated via chelation of the intracellular labile iron pool (LIP) [72]. Due to its hydrophobicity, Ent is able to penetrate the cellular membrane and chelate the LIP, thus limiting the supply of iron whose redox activity are needed to fuel neutrophil oxidative responses [72]. One such response is the release of DNA to form the bactericidal web-like neutrophil extracellular traps (NETs), which is dependent upon MPO, NADPH oxidase and ROS [77]. By inhibiting NETs formation, Ent could confer further protection to bacteria against the antimicrobial activity of neutrophils (Fig. 2) [72]. Other siderophores including pyoverdine, DFO and ferrichrome, also mitigated neutrophil ROS and NETs to a lesser extent than Ent in vitro, further indicating that iron-chelation may explain their inhibitory effects [72]. However, another study found that DFO could slightly increase NETs formation via iron chelation [78]. Such disparity in these outcomes suggest a possible involvement of counter-regulatory responses from neutrophils against weaker iron chelators, such as DFO [36], when compared to Ent. Further studies are certainly warranted to investigate whether siderophores can also impact other types of immune cells, especially macrophages which are rich in iron.

Therapeutics

The increase of in-depth knowledge on siderophores and the innate immune response has led these iron chelators to be exploited to treat iron-overload disorders (e.g., DFO), gut inflammation (e.g., E. coli Nissle), infections and cancer (e.g., ferrichrome) [79, 80, 81, 82, 83]. Specifically, the understanding of the antagonistic relationship between siderophores and Lcn2 has allowed these iron chelators to be utilized for both their canonical and noncanonical functions in various therapeutics. This final section of the review will discuss the future outlooks of utilizing siderophores in medicine.

Sideromycins: A Trojan Horse for Antibiotic Delivery

The emergence of multidrug-resistant bacteria is a major concern, where conventional antibiotics are now deemed largely inadequate as treatments. To counter such a threat, many therapeutics have begun to explore the feasibility of targeting siderophore-mediated iron acquisition to selectively deplete iron-dependent pathogens. One instance is the use of gallium, as a competitive antagonist to iron for siderophores, to hamper metabolic processes that are iron dependent [84]. Another example is to exploit sideromycins, which are antibiotics covalently attached to siderophores, akin to a “Trojan Horse,” which can be tailored to target specific Gram-negative or Gram-positive bacteria [85]. There are two known naturally occurring hydroxamate-type sideromycins: albomycin, a derivative of ferrichrome that inhibits seryl-t-RNA, and salmycin, a derivative of ferrioxamine (holo-DFO) that inhibits a key step in bacterial protein synthesis [85].

In principle, sideromycins have a limited range of susceptible bacteria, whereby their efficacy is dependent on whether the bacteria express the corresponding recep tor(s) to internalize them. For instance, albomycin targets only bacteria expressing the ferrichrome uptake system (e.g., FhuA, FhuD, FhuB) for its transport into the cytoplasm [85]. Nevertheless, such a feature is a necessary advantage in the effort to design antibiotic conjugates that can be delivered to specific groups of bacteria while leaving the rest unharmed. Efforts to produce synthetic sideromycins have been ongoing, many of which have been shown to have greater than 100 times bactericidal activity, when compared to their base antibiotic compound [85]. However, considering the structural mimicry between sideromycins and siderophores, it would be interesting for further research to explore whether sideromycins are capable of being sequestered by Lcn2 and tear lipocalin.

Beyond Iron: Siderophores for Multitarget Therapeutics

Recent studies have discovered noncanonical functions for the chelators, including sequestering other noniron metals such as zinc and copper [86, 87]. Certain siderophores also display the capacity to chelate radioactive metals, i.e., lanthanide and actinide, and act in concert with Lcn2 to facilitate the transport and elimination of such toxic elements [88]. Intriguingly, one study demonstrates that the siderophore-Lcn2 complex could be engineered as a vehicle to deliver radioisotopes into tumors for imaging and radiotherapy [89]. Such findings herald the prospect of engineering siderophores and Lcn2 into multitarget therapeutics capable of treating diseases associated with iron, zinc, copper or heavy metal poisoning, and also cancers. These siderophores could also be coupled with a mutant Lcn2 engineered to evade renal reabsorption [90], thus increasing the efficiency of the complex in eliminating unfavorable elements/biomolecules from the host. Future studies would benefit from studying how a presumably antagonistic relationship between siderophores and Lcn2 could be harnessed into a therapeutic partnership and how this interplay could impact both the host and bacteria.

Anticalins: The Emerging Stealth Lipocalins

Synthesizing a siderophore-based vaccine that is capable of inducing the adaptive immune system to generate anti-siderophore antibodies are interesting avenues for future study [91]. However, an emerging alternative to siderophore antibodies are anticalins, which are molecules “generated by combinatorial design from natural lipocalins” [92, 93], including Lcn2 and tear lipocalin. Lipocalins were an initial target for protein engineering due to their similarities in binding sites with antibodies, where immunoglobulins comprise of six hypervariable loops compared to lipocalins that have a central β-barrel supported by four variable loops that can be modified to recognize small molecules [92, 93]. The use of the recombinant/synthetic DNA technology has already led to the generation of anticalins against human CTLA-4 (cyto­toxic T-lymphocyte-associated antigen 4) [94], VEGFR-3 (vascular endothelial growth factor) [95], the Alzheimer Aβ (amyloid β) peptide [96], and various others (as reviewed in [92]). Future studies should explore whether the generation of engineered lipocalins could sequester stealth siderophores and/or hinder c-di-GMP's antagonizing properties. Hence, anticalins could start a whole new game in the iron tug-of-war between host and bacteria.

Disclosure Statement

The authors have no conflicts of interest to declare.

Funding Sources

M. Vijay-Kumar is supported by R01 grants from the Nation al Institutes of Health (NIH; grant numbers DK097865 and CA219144).

Author Contributions

R.G. drafted and prepared the manuscript and designed the table and figures. B.S.Y and M.V.-K edited and contributed scientific prospects to the manuscript.