Mechanisms of iron and haem transport by Listeria monocytogenes

Abstract

Listeria monocytogenes, the causative agent of listeriosis, is a virulent foodborne Gram-positive bacterial pathogen, with 20–30% mortality. It has a broad ability to transport iron, either in the form of ferric siderophores, or by extracting it from mammalian iron binding proteins. In this review we focus on the mechanisms of ferric siderophore and haem transport into the listerial cell. Despite the fact that it does not synthesize siderophores, L. monocytogenes transports ferric siderophores in the wild environment by the actions of cytoplasmic membrane ABC-transporter systems. The bacterium acquires haem, on the other hand, by two mechanisms. At low (nanomolar) concentrations, sortase B-dependent, peptidoglycan-anchored proteins scavenge the iron porphyrin in human or animal tissues, and transfer it to the underlying ABC-transporters in the cytoplasmic membrane for uptake. At concentrations at or above 50 nM, however, haem transport becomes sortase-independent, and occurs by direct interactions of the iron porphyrin with the same ABC-transporter complexes. The architecture of the Gram-positive cell envelope plays a fundamental role in these mechanisms, and the haem acquisition abilities of L. monocytogenes are an element of its ability to cause infectious disease.

Keywords::

• ABC transporter
• bacterial membrane
• iron transport
• mechanism
• membrane structure

Introduction

This review article considers the mechanistic biochemistry of membrane iron transport by Gram-positive bacteria, among which Listeria monocytogenes is prototypic. Unlike Gram-negative bacterial cell envelope iron uptake systems, which became understood from extensive in vivo and in vitro biochemical analyses over many years, Gram-positive iron transporters were primarily identified only over the past decade from bioinformatic interpretation of genomic data. The in silico approaches created a catalog of potential transport systems, that were subsequently subject to molecular biological and biochemical characterizations. This different chronology of experimental approaches, and in some cases inaccurate genomic annotations, led to misunderstandings of the underlying transport mechanisms. We endeavor to summarize and explain listerial iron acquisition systems, including currently proposed theories of how iron complexes traverse the Gram-positive cell wall. The functions of the transporters relate to the structural features of their membrane systems, and we devote attention to this aspect. In this sense it is germane that our understanding of the Gram-positive cell surface remains rudimentary. Whereas Gram-negative cells have an outer membrane bilayer in which solute transporters reside and interact with the environment, a matrix of peptigoglycan (PG) envelops Gram-positive bacteria, embedded with covalently bound proteins. With regard to the transport of haem (Hn), a focal point of our paper, these surface proteins constitute its initial binding site and are proposed to transfer the porphyrin to ABC-transporters in the underlying cytoplasmic membrane. This postulated process of sequential exchange between or among iron-binding proteins downward through the PG layer distinguishes Gram-positive bacterial transport systems; at the cytoplasmic membrane stage the membrane transporters are comparable to those of Gram-negative cells. We consider the underpinnings and consequences of such proposed mechanisms with regard to their thermodynamic, kinetic and structural features.

A second point of interest in this topic is that bacterial membrane iron transport relates to the widespread pathogenesis of Gram-positive organisms, among which L. monocytogenes is prototypic of a group that includes Staphylococcus, Streptococcus, Bacillus and Corynebacteria. The extent of mortality from these microbes is already high and it is increasing: Deaths from staphylococcal infections alone are about 10,000 and 4,000 per year in the US and UK, respectively. The known relationship between membrane iron transport and the pathogenesis of these organisms is of interest to human health; we will consider this topic more fully elsewhere (Klebba et al. Submitted).

Bacterial mechanisms of iron uptake

The ability to acquire iron plays a role in the outcome of bacterial infections (Griffiths 1993, Tai et al. 1993, Furman et al. 1994, Rouquette et al. 1995, Conte et al. 1996, Tsolis et al. 1996, Ward et al. 1996, Cornelissen et al. 1998, Rozalska et al. 1998). Yet, besides this long-standing (Bullen 1974, 1981, 1985, Bullen et al. 2000) correlation, many fundamental mechanistic questions about bacterial iron uptake remain unanswered. At the cell surface, Gram-negative bacteria contain numerous receptors for uptake of ferric complexes that provide the recognition function of the complicated multicomponent active transport systems of which they are a part. These transporters exist for molecules such as ferric siderophores, as typified by the outer membrane (OM) ferric enterobactin-FepA system (Klebba 2004). Gram-positive cells employ fundamentally different iron transport processes, as a result of their different cell envelope architecture. For both Gram-negative and Gram-positive bacteria, the stepwise biochemical reactions of Fe⁺⁺⁺ transit through the cell envelope are incompletely resolved. L. monocytogenes may acquire iron in at least four ways: Uptake of elemental iron, utilization of exogenous ferric siderophores, extraction of iron from host binding proteins like Tf/Lf/Ftn, and removal of Hn from host proteins like hemoglobin (Hb), hemopexin (Hpx) or haptoglobin-Hb (Hp-Hb) complexes.

Listerial iron acquisition

The fact that L. monocytogenes actively proliferates within cells and spreads from cell to cell suggests that it adeptly transports iron in the intracellular environment. Often microbial siderophores perform this function, because they rapidly extract iron from Tf (Konopka et al. 1982, Konopka and Neilands 1984) and Ftn (Tidmarsh et al. 1983), availing the metal for transport by invading cells. However, L. monocytogenes does not elaborate siderophores; neither does it encode siderophore biosynthetic systems in its genome (Glaser et al. 2001), nor do experimental approaches (chrome azurol S/iron(III)/hexadecyltrimethylammonium bromide (Schwyn and Neilands 1987)) reveal siderophore production during iron-deficient growth (X. Jiang, unpublished data). On the other hand, proteins bound to the listerial cell surface function in the uptake of Hn, either free or incorporated into Hb (Jin et al. 2005, Xiao et al. 2011). Staphylococcus aureus (Mazmanian et al. 2003, Torres et al. 2006, Zhu et al. 2008b), Streptococcus pyogenes (Bates et al. 2003, Fisher et al. 2008, Liu and Lei 2005, Zhu et al. 2008a) and Bacillus anthracis (Maresso et al. 2008), encode comparable Hn/Hb acquisition systems, reiterating their overall importance to Gram-positive pathogens. The ability to acquire iron from mammalian proteins is a common attribute of pathogenic bacteria (Cornelissen and Sparling 1994b), as evidenced by the TonB-dependent Tf/Lf receptors of Neisseria (Cornelissen and Sparling 1994a, Gray-Owen and Schryvers 1996, Bonnah and Schryvers 1998, Schryvers and Stojiljkovic 1999), the Hn/Hb-Hp/hHpx uptake systems of Haemophilus (Sanders et al. 1994, Elkins et al. 1995,1998, Jarosik et al. 1995, Stevens et al. 1996, Ren et al. 1998, Morton et al. 1999, Thomas et al. 2001, Jin et al. 2005) and Neisseriae (Stojiljkovic et al. 1997, Richardson and Stojiljkovic 1999, Zhu et al. 2000), and the broad ability of L. monocytogenes to utilize all known iron mammalian iron sources (Figure 1). So for Listeria monocytogenes, a dichotomy exists: In the wild, the bacterium relies on siderophores produced by competing microbes and utilizes them as an external iron supply. In the host gut, the bacterium may still transport siderophores produced by other competing microbes, but in blood, tissues and the nervous system a different iron transport process occurs: Recognition, adsorption and uptake of Hn. Ultimately, both ferric siderophores and Hn iron complexes use cytoplasmic membrane ABC-transporters as their means of internalization into the bacterial cell.

Figure 1. Iron nutrition tests with L. monocytogenes EGD-e. Bacteria were grown in BHI broth, rendered iron-deficient with 1 mM bipyridyl, and plated on BHI agar containing bipyridyl (Newton et al. 2005). Iron-complexes were applied to the paper discs. EGD-e used siderophores Fc, FcA and FxB [250 uM (discs 1, 4, 7, respectively) 50 uM (discs 2, 5, 8) and 5 uM (discs 3, 6, 9), holoTf and holoLf (discs 10, 11; 6 uM), Hb (disc 13; 15 uM) and Hn (disc 14; 100 uM). Utilization of iron from Ftn (disc 12; 2 uM) and ferric enterobactin (discs 16–18) was not observed here, but seen in other tests (Jin et al. 2005, Newton et al. 2005). This Figure is reproduced in color in the online version of Molecular Membrane Biology.

Despite its noted lack of siderophore biosynthesis, L. monocytogenes transports exogenously available ferric siderophores from other organisms (Coulanges et al. 1996, Jin et al. 2005), and various other iron acquisition systems were also reported. (i) An inducible ferric citrate transport system was found in the prototypic strain EGD-e (Adams et al. 1990) as well as (ii) a cell-surface localized reductase that acted on iron-containing catecholamines and siderophores (Deneer et al. 1995). The reductase had general specificity for iron containing molecules (Adams et al. 1990, Deneer et al. 1995, Barchini and Cowart 1996). In addition, L. monocytogenes was proposed to recognize catecholamine-iron complexes as siderophore-like compounds, and then reduce the ferric-catecholamines to release ferrous iron intracellularly (Coulanges et al. 1997, 1998). This ability to utilize neurotransmitter catecholate iron complexes presumably fit with the predilection of the pathogen for nervous system tissues, but no specific surface receptors for such iron forms were ever identified. Instead, the surface reductase was proposed to broadly recognize and reduce different ferric compounds, including ferric catecholamine complexes. These reports on the surface reductase remain indeterminate because it was neither identified by bioinformatic analyses, nor cloned, nor deleted, nor purified and biochemically characterized. On the other hand, more recent bioinformatic and biochemical characterizations of proteins encoded by Fur-regulated chromosomal loci in EGD-e identified other transport systems. (iii) ABC-transporters exist for ferric hydroxamate siderophores and for Hn (Jin et al. 2005, Xiao et al. 2011). Extensive sequence identity exists between the components of these systems and constituents of other known ABC-transporters (Jin et al. 2005). L. monocytogenes acquires from Tf, Lf and Ftn (Hartford et al. 1993, Jin et al. 2005; Figure 1), but by an undetermined process: No orthologs of Gram-negative bacterial Tf receptors were found in listerial genomes. Lastly, L. monocytogenes produces (iv) high affinity, sortase-anchored binding proteins that facilitate uptake of Hn/Hb at low concentrations (Jin et al. 2005, Newton et al. 2005, Xiao et al. 2011; see below).

Competition for iron among siderophores and eukaryotic binding proteins

For several reasons, the ability of neurologically important catecholamines to supply iron to invading microbes is uncertain. Catecholamines like dopamine, epinephrine and nor-epinephrine contain effective (catecholate) groups for iron chelation. They are structurally comparable to DHBA and DHBS, the biosynthetic precursor and degradation product, respectively, of bacterial catecholates [the Gram-negative siderophore enterobactin and its glucosylated form (also called salmochelin) and the Gram-positive siderophore corynebactin (also called bacillibactin)]. Nevertheless, the affinity of monocatecholates for iron (III) is much, much lower than that of bacterial siderophores, for several reasons. The cell cytoplasm is a reducing environment, while extracellular sites like blood, lymph, the bacterial periplasm and exterior are oxidizing environments. In the latter places, free iron (II) relatively rapidly oxidizes to iron (III); the two oxidation states have different complexation preferences. Iron (III) is a hard Lewis acid with a preference for hard Lewis base ligands, like oxygen. The oxygen ligands almost always distribute around ferric iron in a hexadentate complex with octahedral geometry. Accordingly, molecules that inhabit the extracellular environment, including Tf, Lf and siderophores, chelate Fe⁺⁺⁺ in this fashion. Iron (II), on the other hand, prefers soft ligands like nitrogen and sulfur, and may form tetracoordinate tetrahedral (e.g., iron-sulfur clusters) or hexacoordinate octahedral (e.g., Hn) biological complexes. In the competition for iron among microbial pathogens and eukaryotic iron sequestration systems, the relative affinity of the various iron complexation agents (Table I) mainly determines their ability to capture it in the host environment (although noteworthy exceptions exist (Konopka et al. 1982, Konopka and Neilands 1984)). Catecholamines may penetrate the complexation sphere of Fe⁺⁺⁺ (Arreguin et al. 2009, Porter et al. 2010), but this activity is not equivalent to siderophore-mediated iron-scavenging in the human body. The reason is that the multiple chelation ligands in a single siderophore molecule dramatically increase its overall affinity for iron. The preference of iron (III) for hexacoordinate geometry mandates three catecholamine molecules in its coordination sphere. But, complexation by three independent catechol moieties does not increase the overall affinity of the hexacoordinate complex beyond that of the monomeric ligands, because any of these individual catechols may independently dissociate from the metal center and diffuse off in solution. The usually trimeric nature of siderophores, on the other hand, prevents a single ligand from leaving the iron center if it dissociates from the metal: The two other complexation groups remain attached and the third ligand may re-associate, preserving the overall integrity of the chelate. Furthermore, once an individual siderophore ligand penetrates the complexation sphere of iron in Tf, Lf or Ftn, its other two ligands have a high probability of displacing the other protein side chains, thereby capturing the metal. Lastly, the often huge affinity of siderophores for Fe⁺⁺⁺ iron (Table I) acts as a thermodynamic sink that rapidly shifts the ferrous-ferric equilibrium to the right. The upshot of these considerations is that siderophores have much higher affinity and better ability to complex iron than monomeric species like dopamine, epinephrine or norepinephrine. Thus, the trimeric structure of most siderophores confers two advantages relative to their monomeric components or degradation products or catecholamines: It greatly increases the affinity for iron, and it enhances the ability to extract iron from other chelation spheres, like those of eukaryotic binding proteins. Despite these considerations, siderophore activity was observed for both monomeric catechols (2,3-dihydroxy benzoic acid) in Brucella abortus (Lopez-Goni et al. 1992) and catecholamines in Bordetella bronchiseptica (Anderson and Armstrong 2008, Armstrong et al. 2012).

Table I. Affinities of eukaryotic binding proteins and siderophores for iron (III).

Chelator	K^1	Reference
Transferrin	10^23	(Aisen et al. 1978)
Lactoferrin	3 × 10^25	(Mazurier and Spik 1980)
ApoFerrichrome	10^29.1	(Schwarzenbach 1963)
ApoFerrichrome A	10^29.6	(Wayne and Neilands 1975)
ApoFerrioxamine B	10^30.6	(Schwarzenbach 1963)
Aerobactin	10^22.9	(Carrano 1979)
Enterobactin	10^49	(Carrano 1979)
Corynebactin	10^49	(Bluhm et al. 2002a, 2002b)

¹Association equilibrium constant.

The cell envelopes of bacteria

The mechanisms of prokaryotic membrane transport derive from the structures of their respective bacterial cell envelopes. Gram-negative and Gram-positive cell walls were biochemically, immunologically and microscopically characterized over the past half century, especially in the former case, but the past decade of research produced surprises about their overall architecture. The Enterobacteriaceae, like Escherichia coli, Salmonella, Yersinia, Klebsiella, Serratia, Citrobacter and Proteus, as well as related species Vibrio, Neisseria and Haemophilus, possess a trilaminar cell envelope, composed of an inner membrane (IM), outer membrane (OM), and an aqueous space between them, the periplasm. This macromolecular assembly contains the proteinaceous components needed to explain the uptake of metabolic solutes, including sugars, amino acids, nucleotides, vitamins and metals. The majority of Gram-positive bacteria, on the other hand, apparently only have an inner (or cytoplasmic) membrane (CM), surrounded by an assembly of the macromolecular polymer peptidoglycan (PG). Nevertheless, exceptions may exist, like the Corynebacteria (Hoffmann et al. 2008, Zuber et al. 2008, Marchand et al. 2012) and Mycobacteria (Niederweis et al. 2010), that contain some sort of an OM that includes porins (Trias et al. 1992, Schiffler et al. 2007). The cell surface of L. monocytogenes and related Gram-positive species consists of thick sheaths of the PG biopolymer (Dmitriev et al. 1999, Navarre and Schneewind 1999), to which proteins and other molecules, like techoic acid, covalently or non-covalently associate. However, as we will consider below, in several ways this perceived structure of the Gram-positive cell wall fails to fully explain its physiological transport properties.

Implications of PG structure in the cell wall

Besides its contributions to the physical integrity of the bacterial cells (Yem and Wu 1978, Tomasz, 1979, Scheurwater and Burrows 2011), the structural features of PG are suddenly germane to both Gram-negative and Gram-positive bacterial iron transport processes. Listerial uptake of Hn from animal proteins like Hb involves binding proteins that are covalently bound to PG. For over 40 years the murein polymer was envisioned to surround prokaryotic cells with its glycan [(N-acetyl glucosamine-N-acetyl muramic acid)n; (NAG-NAM)n] chains parallel to the cell surface, and perpendicular to the long axis of the bacterial rod. Microscopically observed PG in the periplasm of Gram-negative bacteria or on the surface Gram-positive bacteria was surmised to contain either a few, or as many as 50 or more of these parallel layers, respectively. PG with glycan chains arranged parallel to the membrane surface was envisioned to assemble and polymerize with a layered architecture. However, an alternative model exists in which the glycan chains organize perpendicular to the plane of the cell membranes (Dmitriev et al. 1999, 2003, 2005). According to this postulate, called the vertical scaffold model, the pentapeptide of PG extends from (NAG-NAM)n backbone in a triskelion pattern that crosslinks into hexagonal cells, creating a honeycomb-like polymer around the bacterial cell (Figure 2).

Figure 2. Models of PG structure. Left: In (A) the layered model (Holtje 1998) predicts oligosaccharide chains perpendicular to the long axis of the bacterial cell and (B) parallel to the cell surface (reprinted with permission from Holtje 1998). (C) Cryotomography of purified, frozen-hydrated Gram-negative bacterial (Caulobacter crescentus) sacculi (reprinted with permission from Gan et al. 2008). The image represents a 10-nm thick Z-slice through the 3-D reconstruction of a purified sacculus. The inset shows enhanded magnification to show glycan moieties (gl). Right (images reprinted with permission from Meroueh et al. 2006): the vertical scaffold model (48, 100), showing the NMR-determined structure of a glycan chain and attached oligopeptide side chains (D), that lead to a hexagonal polymerization (E), creating cells of defined size through the cell wall (F) when the polymer assembles with the NAGNAM chains perpendicular to the cell surface (G). This Figure is reproduced in color in the online version of Molecular Membrane Biology.

PG contains a repeating disaccharide of β-1,4-linked N-acetylglucosamine (NAG) and N-acetylmuramic acid (NAM), to which oligopeptide chains attach (by amide linkage to the lactyl carboxylic acid group of NAM) and crosslink the glycan strands. Variation exists among different species in the composition of the oligopeptide moiety (in L. monocytogenes, it is the tetrapeptide L-Ala-D-Glu-m-Dap-D-Ala; (Kamisango et al. 1982)), the length of glycan polymers, the nature and extent of their crosslinking, and the overall thickness of the PG component. In E. coli, the periplasmic PG layer between the IM and OM is very thin (Labischinski et al. 1991, Matias et al. 2003; for review see Vollmer et al. 2008, Vollmer and Seligman 2010), whereas in Gram-positive cells PG forms a dense outer wall, 2- to 10-fold thicker (Matias and Beveridge 2007). Unexpectedly, the orientation of glycan chains in PG architecture has come in question, in that the layered model (also called the planar network model (Holtje 1998, Koch 2000) predicts them parallel to, whereas the vertical scaffold model (Dmitriev et al. 2003, Meroueh et al. 2006)) predicts them perpendicular to, the cell surface. One ramification of the vertical scaffold model is the formation of a matrix of PG cells with defined, repeating dimensions enveloping the bacterial cells (Meroueh et al. 2006). In this new context, the PG layer may create a previously unappreciated physical architecture that facilitates both Gram-negative and Gram-positive iron acquisition (Kaserer et al. 2008, Xiao et al. 2011; Figure 3). Especially because cell envelope proteins of both prokaryotic families covalently or non-covalently associate with PG, its architecture has implications about iron transport mechanisms.

Figure 3. Space-filling depiction of the PG matrix in the vertical scaffold model. The image in Figure 5 was rotated to show the honeycombed PG polymer (brown) from the cell exterior (100). The pores (20–50 Å) of the polymer allow permeation of small molecules like ferric siderophores and Hn, which the lipoprotein binding proteins of ABC transporters adsorb for uptake. Note the similarity to the arrangement of glycans in the cryotomographic section shown in Figure 2. This Figure is reproduced in color in the online version of Molecular Membrane Biology.

Figure 4. High resolution AFM image of the staphylococcal cell wall. The image shows a network of fibers with large empty spaces. Bar = 50 nm. Reprinted with permission from Touhami et al. (2004).

The demonstration of three-fold symmetry in the side-chains attached to NAM was a key discovery that lent support to the vertical scaffold model of PG (Meroueh et al. 2006). The solution structure of synthetic NAG-NAM oligomers, determined by NMR, showed three- fold symmetry in the lactyl-oligopeptide component of NAM, intimating the formation of hexagonal cells within the polymer. This finding refuted previous suggestions of two-fold symmetry, or 90 degree angles, etc. (Holtje 1998, Koch 2000) in the disposition of the oligopeptide component: the NMR results did not concur with such structures. Furthermore, findings from several crystallographic studies confirmed the three-fold symmetry of peptides attached to the glycan chains of PG (Cho et al. 2007, Perez-Dorado et al. 2007, 2010): In these cases, the same framework of the tetrasaccharide occurred in PG bound to cell envelope proteins, vindicating its triskelion structure and its right-handed helical arrangement (Meroueh et al. 2006). Therefore, convincing evidence supports the scaffold model. From a conceptual standpoint, a vertical orientation of the sugar oligomers offers physiological and biochemical advantages. L. monocytogenes encodes over 100 anchored cell envelope proteins (Popowska 2004), and the organization of PG associated proteins may physically promote a biochemical pathway, or a particular mechanism of solute transfer among such PG-bound components. Finally, the cases of Gram-negative and Gram-positive cell walls may be different. Young summarized the situation (Young 2006), and argued that opportunities exist for several orientations of the PG polymer in the cell wall.

Active iron transporters in the OM and IM of Gram-negative cells

As ferric siderophores traverse the Gram-negative cell envelope they first bind to high affinity OM receptor proteins. Because environmental concentrations of ferric siderophores are rather low, usually not exceeding micromolar levels, the avidity of OM proteins for their ferric siderophore ligands determines the overall efficacy of the iron uptake process. The crystal structures of E. coli FepA and FhuA (Ferguson et al. 1998, Locher et al. 1998, Buchanan et al. 1999) revealed a unique structural fold shared among OM siderophore receptors. Biochemical determinations of their properties (Newton et al. 1999, Cao et al. 2000, Scott et al. 2001, Annamalai et al. 2004) showed that a subnanomolar affinity for ferric enterobactin and ferrichrome, respectively, confers both efficiency and specificity to iron acquisition. It's noteworthy that this high affinity usually confers monospecificity of OM transporters for particular ferric siderophores, and as a result different Gram-negative bacterial species contain multiple surface receptors for the multitude of microbial iron chelators that may exist in their microenvironment. After passage through the OM, ferric siderophores enter the periplasm, where binding proteins (with slightly lesser affinity) adsorb (Sprencel et al. 2000) and deliver them to other transporters in the IM. Several types of IM uptake systems exist (Kuan et al. 1995) but ferric siderophores presumably all enter the cytoplasm through ABC-type permeases (Braun and Cole 1983, Braun et al. 1998). This three-stage transport process describes the uptake of most solutes by Gram-negative bacteria, although many variations on this theme are known (Nikaido and Vaara 1985, Saier et al. 1999, Saier 2000, Dassa and Bouige 2001, Aboulwafa et al. 2003, Zhang et al. 2003, Dassa 2011). For ferric siderophores the periplasmic binding protein is essential to the efficiency of the overall solute transport process. Without it, accumulating iron complexes leak back, or are excreted out of the periplasm, through TolC (Newton et al. 2010).

Sortase-independent and dependent uptake pathways in Gram-positive cells

Without an OM, and hence, in the absence of high affinity ferric siderophore receptor proteins, how do Gram-positive bacteria acquire sufficient quantities of iron to meet their metabolic needs? Either the cytoplasmic membrane ABC transporters of Gram positive bacteria act as the initial binding site for iron-containing molecules in the environment, or the iron acquisition process of Gram positive organisms entails some type of high affinity iron binding proteins within the PG surface layer. Recent findings demonstrate that L. monocytogenes and its relatives employ both transport mechanisms.

Sortase-independent iron transport.

In L. monocytogenes an iron uptake process was described that utilizes a putative non-specific surface reductase (Deneer et al. 1995, Coulanges et al. 1997). The challenge to this pathway is that the relevant membrane protein(s) remain(s) unidentified. The listerial genome contains 25 or more potential membrane oxidoreductases (Glaser et al. 2001), making the identification of the postulated enzyme a daunting task. For instance, site-directed deletion of the annotated reductase lmo1961 in the fhu operon had no effect on ferric siderophore or Hn uptake (Jin et al. 2005, Xiao et al. 2011). So, at present an iron reductase remains undiscovered. Conversely, ABC-transporters were identified and experimentally demonstrated in the listerial CM for uptake of ferric siderophores (Sebulsky et al. 2000, Jin et al. 2005). That is to say, that the fhuBCDG locus, located adjacent to fur at 2.031 Mb in the EGD-e genome, is responsible for the uptake of the hydroxamate iron complexes ferrichrome (Fc), ferrichrome A (FcA) and ferrioxamine B (FxB). Site-directed deletion of genes within this Fur-regulated region completely eliminated the ability to acquire iron from hydroxamate ferric siderophores, and the locus does not encode additional potentially PG-bound proteins. These data confirm that L. monocytogenes and related species directly acquire ferric siderophores from their environments by the activity of cytoplasmic membrane permeases alone. The binding protein, FhuD, of the Fhu membrane permease system manifests a roughly 45 nM affinity for its preferred substrate, FxB, and lower affinities for Fc and FcA (Xiao et al. 2011). The transport K_M values for these hydroxamate iron complexes are about 10 nM; rates for these iron complexes are approximately 25 pMol/10⁹ cells/min (Jin et al. 2005; Table II).

Table II. Thermodynamic and kinetic parameters of ferric siderophore and Hn transport in L. monocytogenes. Binding K_D (nM) and capacity (Cap; pMol/10⁹ cells), and transport K_M (nM) and V_max were determined from studies with [⁵⁹Fe]-Hn.

Transporter/protein	Substrate	KD	Cap	KM	Vmax	Reference
FhuCDGB^1	Fc,FcA FxB	ND^4	ND	10	24	(Jin et al. 2005)
FhuD^2	FxB	45	NA^4	NA	NA	(Xiao et al. 2011)
Unknown^1	FeCrn	ND	ND	10	1.2	(Jin et al. 2005)
HupCDG^1	Hn	10	25	11	4.5	(Xiao et al. 2011)
HupD^3	"	26	NA	NA	NA	"
Hbp2/HupCDG^1	"	2.5	105	1	23	"
Hbp2^2	"	12	ND	NA	NA	"

¹Experiments performed with live bacteria of strain EGD-e.

²Experiments performed with purified binding proteins.

³Experiments performed with live bacteria of strain EGD-e ▵hbp2.

⁴ND: no data; NA: not applicable.

These results have some unexpected ramifications. First, the ability to transport FcA (Figure 1) is novel. Gram-negative bacteria readily transport the neutral compound Fc, but they are unable to transport its anionic (3^-) relative FcA (Wayne and Neilands 1975). Thus, the listerial transporter does not discriminate the two compounds, and recognizes and internalizes hydroxamate ferric siderophores by different biochemical attributes than the Gram-negative bacterial receptor protein. A practical implication of this finding is that apoFcA is inappropriate for iron limitation of L. monocytogenes in nutrition tests (Newton et al. 2005). Secondly, the selectivity of the ferric hydroxamate transport system for FxB concurs with the fact that L. monocytogenes is a saprophytic organism that is widespread in nature: FxB originates from the genus Streptomycetes, which contains Gram-positive bacteria predominantly found in soil and decaying vegetation. The ability to transport this iron chelate is therefore of great utility to L. monocytogenes in wild environments.

Just as for ferric siderophores, L. monocytogenes encodes a CM ABC-transporter system for Hn (HupDGC) that functions during its pathogenesis of animal hosts (Jin et al. 2005, Xiao et al. 2011). Hn directly binds to this Fur-regulated membrane transport system, that resides at 2.499 Mb on the EGD-e chromosome, with K_D = 10 nM. The transport K_M = 15 nM and V_MAX = 23 pMol/min/10⁹ cells for this system (Table II). It is important to note that no obvious barriers exist to Hn diffusion through the PG layer of the Gram-positive cell wall, especially in the vertical scaffold model (Figure 3), and that microscopic (Matias and Beveridge 2007) and biochemical (Demchick and Koch 1996) evidence confirms this inference (Figure 4). The Hn ABC-transporter also illustrates a failing of the genomic approach to identification of membrane transport systems. The original annotation of the EGD-e genome (Glaser et al. 2001) identified three Fur-regulated, putative Fc transporters, at 2.274, 2.031 and 2.499 Mb. However, experimental analyses of these systems by site-directed deletions, cloning of their structural genes and purification of the proteins they encode revealed that only one, at 2.031 Mb, primarily functions in ferric hydroxamate uptake. The ABC-transporter at 2.499 Mb (lmo2431–2429) preferentially transports Hn (it was subsequently designated hupDGC), and the function of the putative ABC-transporter at 2.274 (lmo2182–4), closely linked to srtB, remains unknown (Jin et al. 2005). Deletion of this locus did not affect ferric hydroxamate, ferric catecholate or Hn acquisition. So, the original annotations of the biochemical functions of two of the three loci were in error, and despite the passage of 7 years, remain uncorrected in both ListiList (http://genolist.pasteur.fr/ListiList/) and NCBI (http://www.ncbi.nlm.nih.gov/gene).

Figure 5. Model of the listerial PG, based on the vertical scaffold model. A schematic drawing of the listerial cell wall, from the crystallographic coordinates of homologous constituents, shows the direct permeation of ferric siderophores (Fc; ferrichrome) and Hn to CM ABC-transporters (red arrows), and sortase-dependent uptake of Hn from Hb through PG-anchored binding proteins (blue arrows). This Figure is reproduced in color in the online version of Molecular Membrane Biology.

Sortase-dependent Hn acquisition.

The listerial Hn uptake system is novel relative to the listerial ferric siderophore transport process, in that at low substrate concentrations additional proteins, anchored to external PG, bind Hn with higher affinity to begin uptake. A class of novel enzymes called sortases (Schneewind et al. 1993, Navarre and Schneewind 1994, 1999) process secreted proteins in Gram-positive bacteria by concomitantly cleaving off their C-termini and anchoring their N-termini to PG. Sortases cut their target proteins after T in the generic sequences LPxTG (sortase A) or NPxTN (sortase B), and subsequently form a peptide bond between the α-carboxyl of the Thr or Ser in the cleavage site of the secreted protein and the δ- or ε-amino groups of diaminopimelic acid or Lys side chains, respectively, in the pentapeptide of PG (Cossart and Jonquieres 2000, Dhar et al. 2000, Lebrun et al. 1996). Consequently, the extracellular layers of the Gram-positive bacterial cell envelope become permanently decorated with secreted proteins, including Hn-binding proteins (ranging in size from 25–100 kDa). These PG-associated binding proteins initiate the transport of Hn/Hb in the host environment (Jin et al. 2005, Torres et al. 2006, Hammer and Skaar 2011, Xiao et al. 2011). Yet, it remains unclear how binding proteins anchored within the PG facilitate the Hn acquisition process. Once bound to such proteins, no obvious driving force exists to transfer Hn/Hb to CM permeases. We will consider this question below.

Surface proteins anchored within PG create previously unknown physiological functions of the cell envelope. Up to six different classes of sortases exist; L. monocytogenes encodes sortases in classes A and B (Spirig et al. 2011), that catalyze the linkage of proteins to PG. For example, sortase A (SrtA) recognizes the LPxTG in internalin (Lebrun et al. 1996, Dramsi et al. 1997) and attaches it to the PG matrix. This widespread mechanism (Ton-That et al. 1999, 2000) relates to the function of PG-associated proteins in bacterial virulence, as evidenced by the attenuation of srtA mutants in L. monocytogenes (Bierne et al. 2002), S. aureus and S. pyogenes (Mazmanian et al. 2000, Kharat and Tomasz 2003, Lee and Boran 2003). A second, iron acquisition-related sorting mechanism in the same species involves the srtB gene product (Autret et al. 2001, Bierne et al. 2004). In L. monocytogenes SrtB anchors Hbp1 (the product of lmo2186) and Hbp2 (the product of lmo2185) to PG, probably through the motifs NPKSS and NAKTN, respectively (Bierne et al. 2004, Mariscotti et al. 2009). However, it is mechanistically relevant (see below) that a substantial quantity of these Hn-binding proteins escapes to the extracellular milieu, where they presumably act as hemophores. In S. aureus, sortase B targets another protein implicated in iron uptake, IsdC, through the anchoring domain NPQTN (Mazmanian et al. 2002).

As discussed above, at micromolar concentrations Hn traverses the cytoplasmic membrane by the direct action of ABC-transporters, but at lower (nanomolar) external concentrations a cell surface receptor protein facilitates its uptake by initially binding the iron porphyrin. Such binding proteins anchored to the external cell wall were predicted (Mazmanian et al. 2002), and later shown to extract the iron porphyrin from Hb (Jin et al. 2005, Newton et al. 2005, Nobles and Maresso 2011, Xiao et al. 2011). At low concentrations the sortase-anchored, PG-attached binding proteins are required for efficient Hn transport. So, in a broad sense Gram-positive bacterial Hn/Hb uptake systems encompass both surface receptors and CM transport proteins. Furthermore, besides their sortase-mediated attachment to the PG polymer, the escape or secretion of the same Hn/Hb binding proteins to the external environment (Newton et al. 2005) allows them to act as ‘hemophores', that scavenge free Hn (Newton et al. 2005, Maresso et al. 2008, Xiao et al. 2011). Nevertheless, listerial Hn binding proteins only facilitate Hn uptake when when the bacteria face extreme iron deprivation. At such low (nanomolar) concentrations Hbp2 initiates acquisition by recognizing and adsorbing Hn (Xiao et al. 2011). On the basis of its structural similarity (see below) Hbp1 may similarly participate, but this possibility was not experimentally demonstrated. The presence of PG-anchored Hbp2 in the cell wall greatly enhances the utilization of externally-supplied Hn-Hbp2 (Xiao et al. 2011). Once bound to Hbp1/2, Hn presumably passes from the exterior surface to the CM along a sequential network of the binding proteins, that deliver the iron complex to an ABC transporter complex in the CM bilayer (Liu and Lei 2005, Nygaard et al. 2006, Pilpa et al. 2009). The Hn-binding proteins and the underlying CM ABC transporters have nanomolar affinities for the iron porphyrin (Table II). So, at high external concentrations, the CM permeases may directly adsorb Hn, and at low external concentrations they receive it by unknown mechanisms from the external network of binding proteins, in both cases ultimately internalizing Hn as a result of ATP hydrolysis (Jin et al. 2005, Nygaard et al. 2006; Figure 5).

Hn-binding NEAT-domain proteins

The listerial cell envelope proteins of interest to the Hn/Hb binding reaction are Hbp1 and Hbp2. Homologous proteins exist in Staphylococcus aureus (IsdA, B, C and H), Streptococcus pyogenes (Shr, Shp) and in Bacillus anthracis (IsdX1 and X2; Maresso and Schneewind 2006). This family of Gram-positive bacterial antigens contains diverse overall architectures (Figure 6), but all its constituents have structurally conserved, approximately 125-residue regions (NEAT domains; Andrade et al. 2002) that tightly complex Hn. NEAT domains were so designated from bioinformatic analyses of Gram-positive bacterial genomes, that identified their structural genes and localized them near Fur-regulated loci encoding iron transporters (Andrade et al. 2002). As noted, NEAT-domain-containing proteins either anchor to the PG layer by sortases A and B; (Mazmanian et al. 1999, Ton-That et al. 1999, Bierne et al. 2004), or pass to the external environment (Newton et al. 2005, Maresso et al. 2008, Xiao et al. 2011). Besides this dual localization, different members of the group were reported to target Hn, Hb or Hp as their ligands, so their exact disposition and specificity remains the subject of conjecture.

Figure 6. Hn/Hb binding proteins of Gram (+) bacteria. (A) The srtB region of the L. monocytogenes chromosome contains proteins with either sortaseB (purple loci) and sortaseA (light purple loci) recognition sites. (B) The full-length sequences of IsdA (354 aa), B (645 aa), C (227 aa) and H (895 aa) of S. aureus show homology to Hbp1 (207aa) and Hbp2 (569 aa) of L. monocytogenes. In a CLUSTALW comparison, the ∼ 125 residue NEAT domains (boxed regions) of IsdA, C, Hbp1 and the two NEAT-domains of IsdB or Hbp2 are aligned (once or more) to the three predicted NEAT-domains of IsdH (highlighted green). (C) Homology in NEAT3 of IsdH. Identical amino acids are highlighted blue; black stars mark identity between IsdB and H; red stars mark identity between IsdA or C and Hbp1 or 2; white stars mark identity between IsdB or H and IsdA or C or Hbp1 or 2. NEAT domains of IsdA, C and NEAT3 of IsdH were crystallographically solved, allowing denotation of their secondary structures (8 β-strands and 1 short α-helix; boxed in colors); Hn-contact residues (and identical a.a in other proteins) are highlighted yellow. IsdB (NEAT2) and H (NEAT3) are 58% identical; IsdA, C, Hbp1, 2 and IsdH (NEAT3) are 49% identical. This Figure is reproduced in color in the online version of Molecular Membrane Biology.

From their polypeptide sequences NEAT-domains fall into in four groups (Andrade et al. 2002). These are speculative categories, in that the majority of their proteins are not yet structurally solved; in some cases the original classifications may require revision. But, the bioinformatic analysis identified a family of paralogous or orthologous protein receptors (Figure 6) containing nearly identical structural domains (Figure 7) that complex Hn (not ferric siderophores, as originally assumed; Andrade et al. 2002), and interact with Hn-containing eukaryotic proteins. X-ray crystallographic or NMR structural determinations confirm complexation of Hn by IsdA (Grigg et al. 2007), IsdC (Sharp et al. 2007), and IsdH (NEAT1 and 3; Pilpa et al. 2009). The extent of sequence identity in Hn-contact residues among 6 such NEAT1-containing proteins is ∼60% (Figure 6), which assures conservation of their basic structural motifs (Ginalski 2006). Other Hn/Hb binding proteins exist in Gram-negative bacteria, typified by the hemophore HasA (Letoffe et al. 2001), but aside from a similar molecular mass (∼20 kDa), they are completely distinct in structural organization from the Isd/Hbp proteins.

Figure 7. Structure of NEAT-domains. Three protein domains (IsdA, isdC, IsdH are blue, yellow and green, respectively), aligned by the MATCH algorithm of CHIMERA (UCSF), are illustrated in ribbon format with ligand contact residues (noted in Figure 6) displayed in stick form, in the absence (A) and presence (B) of Hn. This Figure is reproduced in color in the online version of Molecular Membrane Biology.

In light of the participation of PG-anchored Hn-binding proteins in Hn uptake the question arises: How do external binding proteins achieve an influx of Hn to the underlying ABC transporters in the CM? The Hn/Hb binding proteins are passively suspended in the extracellular matrix, at fixed distances from the bacterial CM, collecting low concentrations of Hn in the environment as a result of their high affinity (K_D = 10 nM) interactions with the iron porphyrin. Once associated with the network of extracellular Hn/Hb binding proteins, that ostensibly decorate the PG matrix like leaves on a tree, no obvious rationale for Hn to progress deeper into the cell wall to the CM permease (Figure 5). Two explanations may provide insight into the question. First, transport rate is a relevant consideration. In environments with only low concentrations of Hn the rate of its direct binding and uptake by CM ABC-transporters may be insufficient to sustain a significant bacterial growth. High affinity Hn binding to a network of abundant, nearby proteins will enhance its concentration near the cell surface, accelerating the rate of its transport by CM permeases. For L. monocytogenes the high affinity binding equilibrium created by PG-associated Hbp1 and Hbp2 may locally increase the amount of Hn in the PG layer, raising its concentration from nanomolar external levels to a more effective range for the Hup permease (K_M = 2 nM, so at 20 nM uptake velocity = V_max). Secondly, attachment to the PG matrix may organize Hn-binding proteins into a sequential pathway that leads to the Hup transporters in the CM (Figure 5). The PG polymer is porous to molecules up to 50 kDa (Demchick and Koch 1996), so it permits the diffusion of small molecules like Hn. High resolution microscopic images show the loosely linked PG matrix around Gram-positive bacteria (Touhami et al. 2004), leaving no doubt about its permeability to molecules as large as 50 k_D (Figure 4). With regard to the different proposed configurations of PG constituents, either orientation of glycan chains explains this property: The vertical scaffold model creates regularly arrayed gaps and aligned cells through the cell wall; the layered model predicts similar, but less organized permeability venues. The important point is that PG creates a framework within which functionally related macromolecules may assemble to form a spatially, temporally and thermodynamically viable pathway from the exterior to the CM. This proposed organization has kinetic advantages, similar to those created by the association of different enzymes together to create a multifunctional complex (e.g., the pyruvate dehydrogenase complex; Roche and Hiromasa 2007). Furthermore, the notion facilitates the proposed sequence of protein-mediated acquisition of Hn from Hb by S. aureus, in which sortase A-anchored proteins IsdA, IsdB and IsdH extract the porphyrin from the globin and convey/transfer it to sortase B-anchored IsdC, which delivers the Hn to the HtsABC permease in the CM (Torres et al. 2006, Grigg et al. 2010).

Perhaps the most important ramification of both PG structural models is that neither poses a barrier to the diffusion of small molecules like ferric siderophores and Hn. The particulars of the two PG schemes, for example layer stratification in the planar network model, and regularly arrayed tunnels or portals to the cytoplasmic membrane in the vertical scaffold model, may provide frameworks for the localization of Hn/Hb binding proteins, leading to vectorial Hn transfer mechanisms. But, aside from conceptual postulates, little or no data currently exists to support these ideas.

The multiplicity of Hn/Hb binding proteins in the staphylococcal cell wall, including sortase A-dependent IsdA, B, and H, and sortase B-dependent IsdC, raises the question of whether a comparable multi-protein uptake system exists in L. monocytogenes? Besides Hbp1 and 2, the presence of two genes that encode LPXTG-containing, and hence potentially SrtA-anchored proteins, in its srtB region (Figure 6) underscores this possibility. Lmo2178 and Lmo2179 contain the sortase A recognition sequences LPSTG and LPHTG, respectively, but they do not manifest convincing sequence homology in the structurally relevant, haem-binding site of the NEAT (II) domains of Hbp1 and Hbp2. In particular, they lack relatedness seen in the ligand contact-residues of other known haem-binding NEAT-domain proteins (Figure 6): Among 11 such residues, only one is conserved in Lmo2178, and 3 in Lmo2179 (data not shown). Furthermore, neither structural gene is Fur-regulated, and biochemical characterizations of a ΔsrtA derivative of EGD-e revealed no deficiencies in Hn uptake, nor any obvious defects in Hb utilization (Xiao et al. 2011). At present Hbp2 is the only sortase-anchored cell envelope protein that was directly implicated in listerial Hn acquisition. The activity of other NEAT-domain or sortase A-dependent proteins in this process remains hypothetical.

Hn homeostasis

Despite the demonstrated function of NEAT-domain proteins in listerial iron acquisition, an alternative or additional explanation exists for the presence of Hn-binding proteins in the PG layer enveloping Gram-positive bacterial cells: Detoxification. Hn is harmful at high concentrations. Its extreme hydrophobicity confers membrane permeability, and high external concentrations may translate into lethal cellular levels, from damage to lipids, proteins and DNA as a result of the generation of reactive oxygen species (Ascenzi et al. 2005). Many bacteria express specific mechanisms for detoxification of Hn. Bacterial cell envelope Hn binding proteins (Nhien et al. 2010, Roden et al. 2012), as well as intracellular Hn-binding proteins (Kaur and Wilks 2007) and superoxide dismutases (Negari et al. 2008) provide protection against Hn toxicity in insect and animal host environments, where tissue damage from pathogens may include RBC lysis, Hb degradation and consequent high Hn levels. Regulation of intracellular Hn is essential to bacterial metabolism, and besides systems for its uptake, a variety of Gram-positive bacteria, including L. monocytogenes (McLaughlin et al. 2012), Corynebacterium diptheria (Bibb and Schmitt 2010), S. aureus (Stauff et al. 2008, Anzaldi and Skaar 2010) and Lactococcus lactis (Lechardeur et al. 2012), also encode membrane transport systems for its export. The listerial Fur-regulated locus frvA, that likely also encodes a Hn-exporter (Mclaughlin et al. 2012), is essential for virulence and for resistance to Hn-mediated toxicity. Therefore, the possibility exists that secreted and externally located, PG-anchored Hbp1/2 may prevent excessive intracellular accumulation of Hn, by binding the porphyrin and thereby preventing its unregulated diffusion through the CM into the cell.

Overview of iron transport through the listerial cell envelope

Combined with insight into PG architecture discussed above, a model of the Gram-positive cell wall emerges with proteins attached at various locations and distances from the CM (Figure 5). In the wild environment ferric siderophores simply diffuse through the outer cell wall PG until they encounter appropriate ABC-transporters in the CM, which serve as high-affinity uptake systems (e.g., FhuDCBG, K_M ≈ 10 nM). The details of the host environment that successful pathogens must overcome are sketchy. What are the pertinent iron- or Hn-containing macromolecules in these sites? The elaboration of Hn binding proteins by L. monocytogenes into the extracellular environment and anchored to the PG matrix surrounding the cell argues for their central role in iron acquisition in vivo. The concentrations of free Hn in tissues, whole blood, serum, and at sites of bacterial infection or abscess are not well defined, which complicates our understanding of the relationship between Hn uptake and microbial infection. L. monocytogenes is beta-hemolytic, so at its sites of infection or abscess at least Hb is present, potentially Hn as well, and therefore Hp and Hpx too. Hn may enter listerial cells by direct permeation through PG to its CM ABC-transporter (HupDGC; K_D ≈ 10 nM), or at lower concentrations follow a binding-protein mediated pathway that begins on the cell surface and passes the porphyrin along a sequential network of binding proteins, secreted or anchored by SrtB. The PG polymer may create a barrier to diffusion of the α2β2 Hb tetramer (mw = 64 k_D), and sortase B-anchored proteins Hbp1 and Hbp2 (as well as the orthologous NEAT proteins IsdA, B, C and H of S. aureus (Torres et al. 2006, Stauff and Skaar 2009) and Shr and Shp of S. pyogenes (Lei et al. 2002, Liu and Lei 2005, Nygaard et al. 2006, Zhu et al. 2008a, 2008b) are proposed to bind the protein in the media or on the cell surface and extract its Hn. Therefore, in the host environment listerial Hn/Hb uptake may occur by two pathways: A sortase-independent route that predominates at Hn concentrations above 50 nM, and a high-affinity, sortase-dependent system that functions in the nanomolar concentration range. The latter system may entail a particular architecture of participant proteins that enables direct, physical interactions among PG-anchored Hn-binding proteins. These associations result in extraction of the porphyrin from Hb, its subsequent directional transfer to the HtsABC CM transporter, and final internalization into the cytoplasm, where the porphyrin either enters metabolic/biosynthetic pools or is degraded to free its iron for other metabolic or regulatory purposes.

In the listerial CM the HupDGC permease internalizes Hn into the cytoplasm (Jin et al. 2005, Xiao et al. 2011), directly or indirectly as described above. It is noteworthy that while individual site- directed deletions in the hupDGC locus and the complete operon deletion drastically decreased Hn utilization by EGD-e, the effect was not absolute. Nutrition tests revealed a small amount of residual Hn-promoted growth of L. monocytogenes, indicating the presence of an alternate, secondary Hn transporter. The same deletions increased LD50 of strain EGD-e in the mouse infection model system (100-fold for ΔhupDGC), whereas Δhbp1, Δhbp2 (Jin et al. 2005, Newton et al. 2005, Xiao et al. 2011) and ΔsrtB (Weiss et al. 2004, Jin et al. 2005, Newton et al. 2005, Xiao et al. 2011) had no effect on bacterial virulence. Together the results suggest that direct uptake by HupDGC constitutes both the primary listerial Hn transport pathway and the primary virulence determinant related to iron acquisition. The biochemical activities of PG-anchored binding proteins play a secondary role, in the sense that they only function at low concentrations. In light of their ability to scavenge miniscule concentrations of Hn in the host, the contributions of these proteins to listerial pathogenesis bears further investigation.

Acknowledgements

The research was supported by National Institutes of Health (NIH) grant GM53836 and National Science Foundation (NSF) grant MCB0417694 to PEK and SMN, and funds from Trellis Biosciences, S. San Francisco, CA, USA. The funders had no role in experimental design, data collection and analysis, decision to publish, or preparation of the manuscript.

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.