Beware of proteins bearing gifts: protein antibiotics that use iron as a Trojan horse


  • Rhys Grinter,

    1. Institute of Infection, Immunity and Inflammation, College of Medical, Veterinary and Life Sciences, University of Glasgow, Glasgow, UK
    Search for more papers by this author
  • Joel Milner,

    1. Plant Science Research Theme, School of Life Sciences, College of Medical, Veterinary and Life Sciences, University of Glasgow, Glasgow, UK
    Search for more papers by this author
  • Daniel Walker

    Corresponding author
    • Institute of Infection, Immunity and Inflammation, College of Medical, Veterinary and Life Sciences, University of Glasgow, Glasgow, UK
    Search for more papers by this author

Correspondence: Daniel Walker, Institute of Infection, Immunity and Inflammation, College of Medical, Veterinary and Life Sciences, University of Glasgow, University Place, Glasgow, G12 8TA UK. Tel.: +44 141 330 5082; fax: +44 141 330 5440; e-mail:


Multicellular organisms limit the availability of free iron to prevent the utilization of this essential nutrient by microbial pathogens. As such, bacterial pathogens possess a variety of mechanisms for obtaining iron from their hosts, including a number of examples of vertebrate pathogens that obtain iron directly from host proteins. Recently, two novel members of the colicin M bacteriocin family were discovered in Pectobacterium that suggest that this phytopathogen possesses such a system. These bacteriocins (pectocin M1 and M2) consist of a cytotoxic domain homologous to that of colicin M fused to a horizontally acquired plant-like ferredoxin. This ferredoxin domain substitutes the portion of colicin M required for receptor binding and translocation, presumably fulfilling this role by parasitizing an existing ferredoxin-based iron acquisition pathway. The ability of susceptible strains of Pectobacterium to utilize plant ferredoxin as an iron source was also demonstrated, providing additional evidence for the existence of such a system. If this hypothesis is correct, it represents the first example of iron piracy directly from a host protein by a phytopathogen and serves as a testament of the flexibility of evolution in creating new bacteriocin specificities.


Iron is essential for most life due to its role as a cofactor in the transport and storage of oxygen and in numerous redox reactions (Lindley, 1996). While abundant, iron is effectively insoluble under aerobic conditions making it the limiting nutrient for microbial life in many environments (Krieg et al., 2009). To overcome this obstacle and to obtain iron in a form available for growth, bacteria produce and secrete a diversity of molecules with strong affinity for ferric iron (Fe3+) or iron-containing compounds. These molecules range in size from small organic acids (citrate) to larger siderophores (catecholate) and proteins (haemophores; Ratledge & Dover, 2000). In Gram-negative bacteria, the outer membrane serves as a permeability barrier protecting the cell from antibiotics, detergents and cell-wall-degrading enzymes (Delcour, 2009). However, the outer membrane bilayer also serves as a barrier to the uptake of iron-scavenging compounds and as such it contains a conserved family of β-barrel receptors (TonB-dependent receptors), which selectively transport iron and other nutrient-containing compounds using energy derived from the proton motive force, through interaction with the TonB–ExbB–ExbD complex (Pawelek et al., 2006). Some bacteria also produce receptors for the import of noncognate siderophores (xenosiderophores), providing an advantage to the microorganisms in a mixed community where the vast majority of soluble iron exists in a siderophore complex (Jurkevitch et al., 1992; Greenwald et al., 2009). The availability of iron can also be a deciding factor in the success or failure of bacterial infection, and consequently, mammalian hosts restrict the availability of iron through the production of iron-binding proteins, transferrin, lactoferrin, haemoglobin and ferritin. Siderophores produced by some pathogens bind iron with ultra-high affinity and so are able to scavenge iron directly from host-binding proteins (Weinberg, 2009). Other bacteria acquire iron directly from these host proteins, either through binding to a cell surface receptor or through the production and secretion of binding proteins (Cornelissen & Sparling, 1994). In plants, the mechanisms employed for withholding and obtaining iron during pathogenesis are less well characterized; however, available evidence shows an important correlation between iron acquisition and virulence (Enard et al., 1988; Lemanceau et al., 2009).

TonB-dependent receptors represent an Achilles' heel in the bacterial outer membrane that is exploited by antimicrobial agents seeking to damage or destroy the cell. An example of such agents is the bacteriocins, a diverse class of protein/peptide antimicrobials produced by Gram-negative bacteria to maintain their ecological niche against closely related competitors (Braun et al., 2002). Depending on their site of action, bacteriocins must traverse at least the outer and often both membranes to reach their target. To cross the outer membrane, many bacteriocins possess a receptor-binding domain that binds with high affinity to a TonB-dependent receptor. This positions the protein on the cell surface, leading to interactions with the periplasmic Tol or Ton complexes that many bacteriocins exploit to facilitate cell entry (Chavan & Riley, 2007; Kleanthous, 2010a,b). In the recently identified bacteriocins, pectocins M1 and M2, from Pectobacterium, the receptor-binding domain consists of a horizontally acquired plant-like ferredoxin protein. Strains of Pectobacterium, which are susceptible to these pectocins, are also able to utilize ferredoxin as an iron source (Grinter et al., 2012), suggesting firstly that Pectobacterium possesses a system for iron acquisition from plant ferredoxin and secondly that these pectocins have evolved to directly parasitize this system for cell entry.

This review focuses on how iron acquisition through TonB-linked receptors, provides an advantage to Gram-negative pathogens during pathogenesis and how bacteriocins, specifically pectocins M1 and M2, have evolved to take advantage of these receptors for cell entry.

Iron scavenging using siderophores

The most common strategy applied by bacteria to acquire iron from their environment is the synthesis and excretion of iron-chelating siderophores. Siderophores are structurally diverse, with almost 500 identified to date and generally consist of a flexible, often peptide-derived scaffold with a number of functional groups for coordinating iron (Krewulak & Vogel, 2008). These functional groups (α-hydroxycarboxylic acid, catechol and hydroxamic acid) possess two oxygen atoms which coordinate ferric iron in a bidentate fashion (Boukhalfa & Crumbliss, 2002). This geometry allows siderophores to bind iron with an exceedingly high affinity at physiological pH. As such, siderophores play a pivotal role in pathogenesis of many bacteria including Pseudomonas aeruginosa and Yersinia sp. (Mossialos & Amoutzias, 2009; Fetherston et al., 2010). After the secreted siderophores have bound iron, they are sequestered by specific TonB-dependent outer membrane receptors and the iron–siderophore complex is imported into the periplasm (Braun & Hantke, 2011). In an interesting parallel to the pectocins discussed later, microcin E482m an antimicrobial peptide produced by Klebsiella pneumonia, gains entry to its target cells through the catecholate siderophore receptor FepA. The microcin undergoes post-translational modification with a trimer of N-(2,3-dihydroxybenzoyl) linked to the C-terminal serine residue by a β-d-glucose. This modification which has been shown to bind iron mimics a catechol-type siderophore and significantly increases the toxicity of the peptide (Thomas et al., 2004; Destoumieux-Garzón et al., 2006).

Iron piracy from host proteins

A number of bacterial pathogens with specific mammalian hosts possess systems for directly obtaining iron from host proteins such as transferrin and lactoferrin, which sequester free iron in the body's extracellular fluids. The most thoroughly characterized of these systems is the transferrin transport system of Neisseria gonorrhoeae and Neisseria meningitidis (Noinaj et al., 2012). Transferrin-binding protein A (TbpA), a 100 kDa integral outer membrane protein and transferrin-binding protein B (TbpB) an 80 kDa membrane anchored coreceptor form the basis of this system. TbpA, a TonB-dependent receptor strongly binds transferrin and acts as the conduit for transport of the liberated ferric iron across the outer membrane; however, it lacks the ability to distinguish between the apo and holo forms of the protein (Moraes et al., 2009). The coreceptor TbpB has a strong affinity for the iron loaded transferrin only and acts synergistically with TbpA, considerably increasing the efficiency of iron import (Anderson et al., 1994). Following binding and extraction of iron, apo-transferrin is released from the complex (Lee & Schryvers, 1988). The importance of this system for fitness is demonstrated by the fact that its inactivation renders N. gonorrhoeae avirulent (Cornelissen et al., 1998).

The majority of iron in a mammalian host is stored intracellularly as haemoglobin (Rohde et al., 2002). As such, haemoglobin and the haem it contains represent an important iron source for invading pathogens (Wandersman & Stojiljkovic, 2000). As a result, pathogenic bacteria commonly secrete haemolysins and cytolysins that lyse host cells and release haemoglobin and other haemoproteins (Krewulak & Vogel, 2008). Uptake of the liberated haem is then achieved by a number of specialized systems, which in Gram-negative bacteria generally consist of a TonB-dependent outer membrane receptor, a periplasmic-binding protein and an ABC transporter (Tong & Guo, 2009). An example of a system, where a cell surface receptor directly acquires free or protein-bound haem, is the two-component HpuA/B system of N. meningitidis, which is evolutionarily and mechanistically related to the transferrin-binding system discussed earlier (Rohde et al., 2002). A second system indentified in a number of Gram-negative bacteria and characterized in the opportunistic pathogens P. aeruginosa and Serratia marcescens involves the TonB-dependent receptor HasR and HasA, a secreted haem-binding protein termed the ‘haemophore’ (Arnoux et al., 2000; Alontaga et al., 2009). Structural and biochemical studies of this system have demonstrated that HasA is able to bind free haem or wrest it directly from host proteins, HasA then binds to HasR to deliver its cargo. HasR is also able to obtain haem directly from the environment in the absence of HasA; however, its presence greatly enhances the efficiency of haem uptake (Krieg et al., 2009). Interestingly, members of the genus Pectobacterium, which possesses a highly lytic mode of infection (discussed later) also possess genes encoding HasA and HasR homologues (Franza & Expert, 2010). While not as abundant as in a mammalian host, haemoproteins still represent a potentially important source of iron for an invading phytopathogen (Ajioka et al., 2006; Espinas et al., 2012).

Bacteriocin cell entry

The outer membrane of Gram-negative bacteria provides a first line of defence against harmful substances, such as detergents and some antibiotics, and enables these bacteria to colonize a variety of different and often hostile environments. Consequently, antibiotics that have evolved to efficiently kill Gram-negative bacteria must exploit weaknesses in this defensive membrane. Small molecule antibiotics tend to be either small enough to diffuse through nonspecific pores in the membrane, which allow diffusion of solutes smaller than 600 Da, or hydrophobic enough to diffuse through the lipid bilayer (Delcour, 2009). Colicin-like bacteriocins, protein antimicrobials typified by the well-studied colicins of Escherichia coli and the S-type pyocins of P. aeruginosa (Michel-Briand & Baysse, 2002; Cascales et al., 2007), are ribosomally synthesized proteins ranging in size from 30 to 80 kDa. Many colicin-like bacteriocins have evolved to exploit TonB-dependent outer membrane receptors to enter the bacterial cell. Structurally, bacteriocins consist of a C-terminal cytotoxic domain and N-terminal domains responsible for binding to a receptor on the surface of and translocation into the target cell (Fig. 1). The cytotoxic domain takes the form of either a nuclease domain that targets DNA, tRNA or rRNA, a pore-forming domain that targets the cytoplasmic membrane or a domain that interferes with peptidoglycan synthesis (Sharma et al., 2009). The receptor-binding domain initiates entry into the target cell by binding with high affinity (nanomolar range disassociation constant) to a specific cell surface receptor (Kleanthous, 2010a,b). In the majority of cases, these receptors are TonB-dependent receptors with a physiological role in the binding and import of iron siderophores or other nutrients, such as vitamin B12 (Braun et al., 1994). Colicins are divided into group A and B colicins based on their requirement for the Tol or Ton systems for cell entry. Tol and Ton are evolutionarily related protein complexes that are anchored to the inner membrane and span the periplasm. They are coupled to the protein motive force, with the Ton system functioning as an energy transducer for the uptake of nutrients through the TonB-dependent outer membrane receptors (Kleanthous, 2010a,b). The function of the Tol system is less well understood; however, mutants deficient in components of the system are more sensitive to EDTA and deoxycholate and it is recruited to the septation apparatus during cell division where it plays a role in stabilizing the outer membrane (de Zwaig & Luria, 1967; Kleanthous, 2010a,b).

Figure 1.

Crystal and domain structure of colicin M, Crystal structure of colicin M as determined by Zeth et al. (2008) (PDB ID = 2XMX), with functional domains delineated as: translocation domain (blue), receptor-binding domain (yellow) and cytotoxic domain (green). This domain structure is representative of colicins identified to date (Zeth et al., 2008).

The translocation domain of colicins facilitates entry by interaction with a component of the Ton or Tol system in the periplasm. A large portion of this domain consists of an inherently unstructured region which reaches the periplasm by threading through the lumen or down the side of an outer membrane porin, or in the case of colicin Ia an additional copy of its receptor. This unstructured region contains a specific epitope, which in the case of group B colicins mimics the TonB box of outer membrane receptors interacting with TonB via β-augmentation (Baboolal et al., 2008; Housden et al., 2010; Jakes & Finkelstein, 2010). The exact mechanisms of how these interactions lead to translocation are yet to be completely understood; however, it is clear that a number of colicins utilize not only the receptors, but also much of the machinery involved in siderophore import.

Pathogenesis of Pectobacterium

The bacterial family Enterobacteriaceae contains many well-studied species which form commensal or pathogenic relationships with humans, including the genera Salmonella, Yersinia, Shigella and Escherichia (Glasner & Perna, 2004). This family also contains a number of phytopathogens including members of the genus Pectobacterium (formerly Erwinia); the causal agent of soft rot and black leg disease. This genus contains species with both broad and restricted host ranges, which cause the above-mentioned diseases in a number of economically important crops including potato, sugar beet and maize (Ma et al., 2007). A key feature of the genus is the production of a range of lytic enzymes during infection which leads to lysis of host cells and a characteristic maceration or soft rotting of host tissues (Pérombelon, 2002). The hydrolysis of pectin during this process provides oligogalacturonides that are utilized by the bacteria as a carbon source, while the associated lysis of the host cells releases intracellular micronutrients such as iron (Expert, 1999). Due to its role in the creation of oxygen radicals via the Fenton reaction and to limit its availability to invading pathogens, the vast majority of intracellular iron in plants is sequestered by haem or iron–sulphur-proteins or the iron storage protein ferritin (Briat, 2007; Briat et al., 2010). Experimental studies into the iron acquisition systems possessed by Pectobacterium, their regulation and role in virulence are limited; however, analysis of data from existing studies along with the genome sequences of a number of strains suggests the genus uses multiple systems for obtaining iron during infection (Franza & Expert, 2010). For example, the genome of Pectobacterium carotovorum SCRI1043 contains a gene cluster for the biosynthesis and transport of the siderophore enterobactin, which has been shown to be regulated by quorum sensing (Bell et al., 2004; Monson et al., 2012). Genes encoding the transport machinery, but not biosynthesis of achromobactin are also present, suggesting it may be utilized as a xenosiderophore (Franza & Expert, 2010). The role of these systems in virulence is yet to be tested and as Pectobacterium can adopt a saprophytic, soil-dwelling lifestyle, iron acquisition during infection may not be their prominent role (Toth et al., 2006). Iron-uptake systems more likely to be involved in virulence are a ferric citrate uptake system and the HasA/HasR system discussed earlier. Plants utilize citrate to transport ferric iron to photosynthetic tissues via the xylem, suggesting uptake of this complex may be important during vascular colonization by the pathogen (Thomine & Lanquar, 2011). As our understanding of pathogensis-related iron-uptake systems in Pectobacterium is still limited, it is quite possible that the genus may have evolved unique mechanisms to obtain iron from its host.

Pectocins M1 and M2

Two bacteriocins Pectocin M1 and M2 from Pectobacterium were recently characterized by our laboratory (Grinter et al., 2012). The cytotoxic domain of these proteins is homologous to that of colicin M, which functions by cleaving the peptidoglycan precursor lipid II (El Ghachi et al., 2006; Zeth et al., 2008; Barreteau et al., 2009; Fig. 1). We identified these proteins bioinformatically based on similarity to colicin M and this similarity was also noted by Helbig et al. (Helbig & Braun, 2011). Due to its low abundance and key role in cell-wall synthesis, lipid II constitutes a common vulnerability among bacteria and is also targeted by a number of peptide-antibiotics (Breukink & de Kruijff, 2006; Schneider et al., 2010). Based on homology to the catalytic domain of colicin M, putative colicin M-like bacteriocins have been identified in a number genera of the γ-proteobacteria (Barreteau et al., 2004).

Pectocin M sequence homology with colicin M is confined to the minimum C-terminal region of colicin M required for cytotoxic activity (Barreteau et al., 2009). Strikingly, the remainder of the protein, which in colicin M consists of a helical receptor-binding domain and unstructured N-terminus, has been replaced through recombination with a plant-like [2Fe-2S] ferredoxin domain with an intact iron–sulphur cluster (Palmer et al., 1967; Grinter et al., 2012; Fig. 2). [2Fe-2S] ferredoxins represent a super family of small (≈100 amino acid) soluble proteins, which contain a single [2Fe-2S] cluster coordinated by four conserved cysteine residues and are predominantly found in the chloroplasts of plants and cyanobacteria (Fukuyama, 2004). Their primary function is to transfer electrons from photo-reduced photosystem I (PSI) to ferredoxin NADP+ reductase, which produces NADPH for CO2 assimilation. They also play a secondary role in distribution of electrons from PSI for assimilation of inorganic nitrogen and sulphur (Fukuyama, 2004; Hirasawa et al., 2009). Ferredoxin's key role in these processes means that it is one of the most abundant iron-containing proteins in photosynthetic organisms (Merchant & Sawaya, 2005; Terauchi et al., 2009). Related [2Fe-2S] ferredoxins and ferredoxin-containing domains are distributed throughout all trees of life from CarE, a [2Fe-2S] ferredoxin involved in carbapenem biosynthesis in Gram-negative bacteria, to adrenodoxins found in vertebrates that facilitate electron transfer from NADPH-dependant ferredoxin reductase to cytochrome P450 (McGowan et al., 1996; Ewen et al., 2011). Phylogenetic analysis shows that the ferredoxin domain from pectocin M is most closely related to plant ferredoxins, indicating that the encoding gene was acquired as a result of horizontal gene transfer, most probably from a host plant (Grinter et al., 2012). Pectobacterium also contains other ferredoxin genes of plant origin that have been implicated in protection from oxidative stress (Sjöblom et al., 2008).

Figure 2.

Homology of pectocin functional domains, homology of ferredoxin domain (red) from pectocin M1 (from Pcc. PC1), pectocin M2 (from Pcb. MBPR1692) and pectocin P (from Pcc. WPP14) with ferredoxin I from Spinacia oleracea and cytotoxic domain to colicin M (from Escherichia coli H299) (green) and pesticin (from Yersinia pestis CO92) (blue).

In our study, we tested the killing spectrum of pectocin M1 and M2 against a number of members of the γ-proteobacteria, finding them to be active only against other strains of Pectobacterium (Grinter et al., 2012). This narrow specificity is typical of bacteriocins, as they bind with a high degree of specificity to their cognate outer membrane receptor (Zamaroczy & Chauleau, 2011). Under nutrient-rich conditions, the activity of the pectocins was weak and only detectable against a limited number of strains; however, the activity of pectocin M1 was significantly enhanced under iron-limiting conditions with inhibition of over 70% of strains tested. These data suggest the receptor responsible for cell entry is widely distributed among strains of Pectobacterium and is strongly regulated by iron availability. Iron-dependent activity is also observed in a number of pyocins, which utilize the receptor responsible for the uptake of the siderophore pyoverdine (Elfarash et al., 2012).

Because of the strong sequence identity between the pectocin ferredoxin domain and a [2Fe-2S] plant-like ferredoxin, we investigated the ability of ferredoxin I from spinach, a catalytically inactive version of pectocin M1 and recombinant human adrenodoxin to interfere with the cytotoxicity of pectocin M1. We found that both the spinach ferredoxin and the inactive pectocin M1 mutant were able to inhibit the cytotoxic activity of pectocin M1. This inhibition suggests competition occurs between the pectocin and the plant-like ferredoxin for the outer membrane receptor responsible for pectocin M cell entry. Adrenodoxin at a concentration well in excess of that used for the plant ferredoxin failed to inhibit activity, demonstrating that this effect is not nonspecifically due to the addition of iron. During this experiment, it was also observed that under iron-limiting conditions the growth of a number of Pectobacterium strains was enhanced by spinach ferredoxin, but not by adrenodoxin. In a strain resistant to pectocin M1, a reciprocal effect was observed where the growth enhancement due to spinach ferredoxin was inhibited by pectocin M1 (Grinter et al., 2012). Analysis of these data leads to the conclusion that Pectobacterium possesses a receptor which specifically binds plant ferredoxin. The ferredoxin's ability to interfere with pectocin M activity and the reciprocal effect where pectocin M interferes with ferredoxin growth enhancement strongly suggest that these proteins interact with the same cell surface receptor.

Pectobacterium ferredoxin piracy model

Based on existing knowledge of systems utilized by Gram-negative pathogens to scavenge iron from host proteins and the data from our study on pectocin M1 and M2, we propose a model for the acquisition of iron from host ferredoxin by Pectobacterium during pathogenesis. In this model outlined in Fig. 3, ferredoxin is sequestered by a specific cell surface receptor or receptor complex, which then either removes the iron–sulphur cluster on the cell surface and releases apo-ferredoxin or imports ferredoxin into the periplasm where it is processed to remove iron. Iron could then be bound by a periplasmic-binding protein and imported to the cytoplasm but its cognate inner membrane ABC transporter (Andrews et al., 2003). This system could be most simply exploited by pectocin M if the entire ferredoxin protein was imported, as a system capable of importing a folded ferredoxin could likely inadvertently also import the colicin M-like cytotoxic domain. However, in systems indentified thus far iron is removed from the protein on the cell surface and independently imported into the cell. If this were the case, the ferredoxin domain of pectocin M may provide only a receptor-binding function, with another part of the protein playing a role in translocation into the periplasm, possibly through interaction with an additional receptor as is the case for most colicins (Fig. 4; Cascales et al., 2007). Interestingly, analysis of existing Pectobacterium genomes reveals an uncharacterized open reading frame (designated pectocin P) which consists of a ferredoxin domain fused to a domain homologous to the catalytic domain of the peptidoglycan degrading bacteriocin pesticin (Fig. 2). This fusion with an unrelated cytotoxic domain with its site of action in the periplasm suggests flexibility in the ability of the ferredoxin domain to mediate translocation of structurally unrelated protein domains.

Figure 3.

Model of iron acquisition from plant ferredoxin by Pectobacterium, (a) outer membrane receptor imports entire ferredoxin into periplasm, where iron is extracted, (b) outer membrane receptor extracts iron–sulphur cluster from ferredoxin on cell surface and imports it to periplasm, releasing apo-ferredoxin.

Figure 4.

Model of pectocin M import by ferredoxin receptor. (a) Outer membrane receptor binds to ferredoxin domain of pectocin M, cytotoxic domain is imported in conjunction with ordinary ferredoxin import. (b) Outer membrane receptor binds ferredoxin domain, pectocin M forms secondary interaction with a receptor facilitating translocation (Pilsl et al., 1993; Barreteau et al., 2009; Helbig & Braun, 2011).


The characterization of pectocin M during a study aimed at identifying novel bacteriocins to combat Pectobacterium-related disease has seemingly identified a novel system which this organism uses to acquire iron form its host. Molecular characterization of this system, through the identification of the import machinery, will provide insight both into the function and evolution of iron piracy in Gram-negative bacteria and the mechanisms by which bacteriocins translocate across the outer membrane. Additionally, investigation of its importance for pathogenesis may allow for the development of strategies to combat Pectobacterium infection in the field.


R.G. is supported by a Kelvin Smith Scholarship funded by the University of Glasgow.