Pathogenic neisseriae have a repertoire of high-affinity iron uptake systems to facilitate acquisition of this essential element in the human host. They possess surface receptor proteins that directly bind the extracellular host iron-binding proteins transferrin and lactoferrin. Alternatively, they have siderophore receptors capable of scavenging iron when exogenous siderophores are present. Released intracellular haem iron present in the form of haemoglobin, haemoglobin–haptoglobin or free haem can be used directly as a source of iron for growth through direct binding by specific surface receptors. Although these receptors may vary in complexity and composition, the key protein involved in the transport of iron (as iron, haem or iron-siderophore) across the outer membrane is a TonB-dependent receptor with an overall structure presumably similar to that determined recently for Escherichia coli FhuA or FepA. The receptors are potentially ideal vaccine targets in view of their critical role in survival in the host. Preliminary pilot studies indicate that transferrin receptor-based vaccines may be protective in humans.
Iron is an essential element, as it serves as a cofactor in key metabolic processes, such as nucleotide biosynthesis and energy production. In iron-containing proteins, the iron is bound directly by amino acids or may be incorporated into haem, a complex of iron with porphyrin (Bridges and Seligman, 1995). Excess iron within cells is stored in the protein ferritin and, in spite of the redox status inside cells, the iron is stored as ferric and not ferrous ions. In the human host, iron bound to haemoglobin within erythrocytes plays a critical role in the transport and exchange of oxygen with tissues throughout the body. The tetrameric haemoglobin molecule consists of four homologous 16 kDa subunits that each bind a single haem moiety.
Iron is transported throughout the body by the serum glycoprotein transferrin (Tf) (Bridges and Seligman, 1995). Human Tf is an 80 kDa bilobed monomeric glycoprotein with a ferric ion and bicarbonate anion binding site present in each lobe. Iron-loaded Tf is bound by transferrin receptors on the surface of cells requiring iron, and the iron is removed through a cyclical process involving receptor-mediated endocytosis. Tf is present in substantial levels (25–44 μM) in serum, but is only a minor component (0.2–1.3 μM) of mucosal secretions. In serum, the iron saturation of Tf is usually 33%, with iron primarily occupying the N-lobe, and the level of iron saturation can be reduced substantially during infection.
Lactoferrin (Lf) is a related iron-binding glycoprotein that is present in mucosal secretions and is released by leucocytes at sites of inflammation. Lf is structurally very similar to Tf but differs in that it is capable of retaining iron under acidic conditions (pH < 6). In contrast to Tf, Lf is involved in a variety of physiological functions other than iron sequestration (Spik et al., 1998). Lf is present at very low levels in serum (3.8–8.8 nM) but is a significant component of mucosal secretions (6–13 μM). Thus, the extracellular milieu in areas throughout the human body has substantial levels of the iron-binding glycoproteins, which effectively reduce the level of free ferric ions below that required to sustain microbial growth.
Iron bound to haemoglobin (Hb) constitutes nearly two-thirds of the total iron in the human body (Bridges and Seligman, 1995), but it is not readily available to pathogens because of its compartmentalization within erythrocytes. Small amounts of Hb (80–800 nM) are found in normal human serum as a result of spontaneous haemolysis, but it is rapidly complexed by the excess quantities (4.7–21 μM) of the circulating heterotetrameric glycoprotein, haptoglobin (Hp), and removed by the liver. Released haem in serum is rapidly complexed by the specific binding protein haemopexin (12 μM) or by excess quantities of albumin and removed by the liver (Sassa and Kappas, 1995).
The sources of iron on human mucosal surfaces available to bacteria are not as well understood. Shedding of dead cells from epithelial surfaces (1 mg iron day−2) would release iron predominantly in the form of ferritin (25% of total body iron), with only trace amounts from haem-containing proteins (0.1% of total body iron) (Bridges and Seligman, 1995). A considerable quantity of dietary iron (18 mg day−1) would be available on mucosal surfaces of the gastrointestinal tract. However, its impact on the mucosa of the nasopharynx is uncertain. Monthly menstrual bleeding in females provides a considerable supply (12–29 mg) of iron to the female genitourinary tract, but the availability of erythrocyte-derived iron on the nasopharyngeal and male genitourinary tract may be limited and episodic at best.
Mechanisms of iron acquisition in the pathogenic Neisseria
Siderophore-mediated iron acquisition
It has been fairly well established that the pathogenic Neisseria do not produce siderophores (West and Sparling, 1985) and thus cannot use this mechanism of iron acquisition during the invasive phase of infection (i.e. in serum and cerebrospinal fluid). However, in the milieu of the mucosal surface, siderophores that are being produced by neighbouring microbes would provide a potential source of iron. Exogenous siderophores have been shown to support the growth of pathogenic Neisseria (West and Sparling, 1985), suggesting that they possess pathways for the uptake of iron–siderophore complexes. Recently, it has been shown that FrpB, an abundant, iron-regulated, 70 kDa protein in pathogenic Neisseria species, is a functional homologue of the Escherichia coli ferric enterochelin receptor (FepA) (Newton et al., 1998). The binding constant of FrpB for enterochelin was significantly lower than that of other enterochelin receptors. This raises the question of whether an alternative, but structurally related, siderophore might be the authentic ligand for FrpB in vivo.
Analysis of the meningococcal and gonococcal genome sequences revealed at least three additional open reading frames (ORFs) that share homology with siderophore receptors (Turner et al., 1998a), suggesting that pathogenic Neisseria species may have a repertoire of siderophore receptors that would be able to use the siderophores released by neighbouring microbes. Comparisons of the specificity and regulation of the meningococcal and gonococcal receptors will be interesting in view of the different mucosal environments that they occupy.
Iron acquisition from transferrin and lactoferrin
Pathogenic neisseriae produce surface receptors capable of binding Tf specifically (Gray-Owen and Schryvers, 1996) as the initial step in iron acquisition from this host glycoprotein. The transferrin receptor consists of two proteins, transferrin-binding protein A (TbpA) and B (TbpB). The genes encoding the receptor proteins, tbpB and tbpA, are arranged in an operon (Legrain et al., 1993) containing promoter elements, suggesting that expression under iron-limiting conditions is mediated by a Fur repressor homologue. It appears that limiting iron is the only signal required for receptor expression (Schryvers and Morris, 1988) and, as there is no evidence of phase variation, the transferrin receptor would be expected to be expressed in vivo during all phases of colonization and infection. Tf was not expected to be an important source of iron on mucosal surfaces because of the relatively low levels reported to be present (Bridges and Seligman, 1995). However, the observation that an isogenic Tbp-deficient mutant was avirulent in a human gonococcal infection model, while the parent strain was virulent (Cornelissen et al., 1998), indicates that there is sufficient Tf on the genitourinary mucosa in vivo to support growth. In view of the levels of Tf present in serum and other body compartments, the Tf receptor is probably important for iron acquisition during the invasive phases of infection. The invariant presence of Tf receptor genes in clinical isolates of meningococci and gonococci provides further support for its essential role in survival in the host.
Pathogenic neisseriae also produce surface receptors capable of binding Lf specifically and acquiring iron from this glycoprotein. The receptor was initially thought to consist of a single receptor protein, LbpA (Schryvers and Morris, 1988), but a second receptor protein, LbpB, has been identified recently (Bonnah and Schryvers, 1998; Pettersson et al., 1998). LbpB shares homology with TbpB, but differs in that it has two distinct regions rich in negatively charged amino acids. The operonic organization of the lbp genes and regulation of expression by iron-limiting conditions parallels the results with the transferrin receptor. Although Lf is present on mucosal surfaces and presumed to be a preferred iron source, several lines of evidence suggest that it is not critical for colonization of mucosal surfaces. First, an Lbp-deficient gonococcal strain, but not its Tbp-deficient derivative, is virulent in a human infection model (Cornelissen et al., 1998). Furthermore, a significant proportion of clinical gonococcal isolates are deficient in lactoferrin utilization, which interestingly seems to result from a deletional event that is being spread horizontally to different strains (Anderson et al., 1998). In this respect, the invariant presence of Lbps in meningococci might indicate that Lf is a more important source of iron on the nasopharyngeal mucosa or may perhaps implicate an additional role for Lbps in meningococcal survival.
Iron acquisition from haem and haemoglobin
Pathogenic neisseriae are capable of using haemoglobin (Hb), haptoglobin–haemoglobin (Hp-Hb) and haem but not haem–haemopexin or haem–albumin as a source of iron for growth (Dyer et al., 1987). They have two distinct surface receptors involved in haem iron acquisition. HmbR is an iron-regulated 89.5 kDa outer membrane protein that mediates iron acquisition from haem and Hb but not from Hb-Hp (Stojiljkovic et al., 1996). The second receptor consists of an 85 kDa outer membrane protein, HpuA, and a 34.8 kDa lipoprotein, HpuB, that is essential for acquiring iron from Hb and Hb-Hp (Lewis et al., 1997).
The level of haem on mucosal surfaces is normally quite low. However, gonococci could be exposed to an abundant supply of haem iron periodically during menstrual bleeding which would presumably be in the form of free Hb, Hb-Hp and various forms of complexed and free haem. It is interesting that gonococci primarily use the HpuAB receptor because of a premature stop codon in the hmbR gene (Chen et al., 1996; Stojiljkovic et al., 1996), perhaps reflecting the prevalence of Hb-Hp as an iron source under these conditions. Phase variation of expression of HpuAB, resulting from slipped-strand mispairing in a poly(G) tract in the hpuA gene (Chen et al., 1998), also seems appropriate considering the episodic nature of the supply of haem iron.
Although the supply of haem in the nasopharynx is probably limited, two haem-requiring pathogens, Porphyromonas gingivalis and Haemophilus influenzae, successfully colonize human oropharynx and nasopharynx respectively. Thus, small amounts of haem may be available as an iron source to meningococci, and they would presumably require the expression of the TonB-dependent receptors, HmbR and/or HpuB, for effective uptake. During the invasive phase of infection, although the supply of haem iron might be relatively small (i.e. > 25 μM Tf versus < 0.8 μM Hp/Hb in serum), a mild degree of local haemolysis would alter the situation dramatically. The haemolysis would increase the concentration of Hp-Hb complexes, exhaust free Hp molecules and make free Hb available for assimilation by bacteria. The expression of both HmbR and HpuAB receptors undergoes phase variation (Lewis et al., 1998; Stojiljkovic et al., 1998). This may provide a mechanism for adapting to variations in the supply of various haem iron sources and/or for limiting the exposure of Hb receptors to the immune system. The pathogenic Neisseria species would be in the intracellular environment for at least part of the infectious process, and little is known about whether they can use any of the intracellular iron stores during this phase. However, the inability of haem biosynthetic mutants to grow within epithelial cells (Turner et al., 1998b) suggests that haem is not readily available during intracellular growth.
Composition and structure of the iron acquisition pathways
Siderophore-mediated iron acquisition
Although siderophore-mediated iron acquisition in pathogenic Neisseria has not been well characterized, there has been considerable progress in understanding this mechanism in other species, which can be readily applied to the systems in gonococci and meningococci (Fig. 1). The outer membrane ferric aerobactin (Ferguson et al., 1998; Locher et al., 1998) and ferric enterochelin (Buchanan et al., 1999) receptor proteins, FhuA and FepA, from E. coli have been crystallized recently and shown to consist of a transmembrane beta-barrel with surface loops that restrict, but do not block, access to the central pore from the external surface. The N-terminal portion of the protein forms a ‘plug’ that fills the beta-barrel from the periplasmic side, with the upper portion contributing to the iron-siderophore binding site. Binding of iron-siderophore by surface loops on the plug is associated with conformational changes on the periplasmic face (Locher et al., 1998) that presumably enhance its interaction with TonB. The TonB–receptor interaction provides the energy required for transport of iron-siderophore across the outer membrane by an as yet undefined process. The iron-siderophore is subsequently bound by a periplasmic iron-siderophore binding protein that is responsible for shuttling the iron-siderophore to an inner membrane transport complex (Fig. 1), which mediates the transport of iron-siderophore across the inner membrane.
Iron acquisition from transferrin and lactoferrin
The Tf and Lf receptor-mediated pathways of iron acquisition appear to have a similar overall organization to the siderophore-mediated pathways with similar outer membrane, periplasmic and inner membrane components and, thus, probably share a similar mechanism of transport (Gray-Owen and Schryvers, 1996). However, the Tf and Lf receptors must accomplish the additional task of removing iron from these high-affinity iron-binding glycoproteins. The demonstration that TbpB-deficient (Anderson et al., 1994) and LbpB-deficient (Bonnah and Schryvers, 1998) mutants are still capable of using Tf/Lf as a source of iron for growth, albeit much less effectively, indicates that TbpA and LbpA are capable of mediating the entire process themselves. This might suggest that the ancestral Tf/Lf receptor may have consisted of a single component, analogous to the siderophore receptors, and that the lipoprotein component was acquired later for increasing the efficiency of the process.
The sequence homology of TbpA and LbpA with siderophore receptors, albeit weak, suggests that they may have a similar overall structure and mechanism of transport. TbpA and LbpA are considerably larger than siderophore receptors (+ 20 kDa), and this is primarily in the size of the predicted surface loops (Pettersson et al., 1994), several of which have been demonstrated experimentally (Prinz et al., 1998). The surface loops are presumably required to facilitate iron removal from Tf/Lf and for interaction with TbpB/LbpB. The removal of iron from Tf may be accomplished simply by inducing conformational changes upon binding, which is supported by gel filtration experiments with the purified proteins in which TbpA alone is capable of removing iron from Tf (Gómez et al., 1998). As the liganding amino acids in Tf are on the surface of the two domains that form the iron-binding cleft, a shift to separate the two domains would reduce the binding affinity for iron (Fig. 2).
TbpB and LbpB are outer membrane lipoproteins presumed to be anchored to the outer membrane by the fatty acyl groups attached to the N-terminal cysteine (Lissolo et al., 1994). TbpB is proposed to play a facilitatory role in the iron acquisition process. Although isogenic TbpB-deficient meningococcal mutants were incapable of growth with 1 μM Tf as a source of iron (Irwin et al., 1993), gonococcal TbpB-deficient mutants demonstrated growth, albeit reduced, in the presence of high concentrations of Tf (Anderson et al., 1994). The surface accessibility of TbpB combined with its preference for the iron-loaded form of Tf (Retzer et al., 1998) suggests that it may serve in the initial capture of Tf. TbpB, like TbpA, can facilitate the release of iron by Tf, but the receptor complex is more effective (Gómez et al., 1998). The demonstration that TbpB binds to peptides on the surface of Tf that essentially wrap around each lobe (Retzer et al., 1999) indicates that TbpB could facilitate separation of the two domains, which is proposed to be responsible for the release of iron from Tf (Fig. 2). The sequence homology between TbpB and LbpB suggests that LbpB may perform a similar function in the lactoferrin receptor, but experimental evidence is currently lacking.
Iron transport across the outer membrane by TbpA or LbpA requires a functional TonB protein (Stojiljkovic and Srinivasan, 1997), and this interaction is presumably similar to the interaction of TonB with siderophore receptors. After transport across the outer membrane, the iron is subsequently bound by the periplasmic iron-binding protein FbpA (Chen et al., 1993), but it is not known whether this involves interaction between TbpA/LbpA and FbpA. The crystal structure of the H. influenzae FbpA (Bruns et al., 1997) and, more recently, the gonococcal FbpA has been determined. The overall structure resembles a single lobe of Tf but differs in the specific liganding amino acids and bound anion (phosphate versus carbonate). The iron–FbpA complex is subsequently bound by an inner membrane FbpB/FbpC complex, which is presumed to be responsible for transport across the inner membrane.
Iron acquisition from haem and haemoglobin
The homology in amino acid sequence and a requirement for functional TonB implies that Hb receptors in pathogenic Neisseria are structurally and functionally similar to siderophore, transferrin and lactoferrin receptors. Both Hb receptors bind Hb, but only HpuAB is capable of binding Hp-Hb and apo-Hp (Chen et al., 1996; Stojiljkovic et al., 1996; Lewis et al., 1997). The binding of substrates by HmbR and HpuAB is a TonB-independent event. Hb receptors are capable of removing haem from Hb before transport across the outer membrane (Stojiljkovic and Srinivasan, 1997). The mechanism by which this is accomplished has not been determined but, similar to what we are proposing for TbpA/LbpA (Fig. 2), it probably involves induced conformation changes in Hb.
The HpuAB receptor resembles the Tf/Lf receptors in some respects, in that it consists of a predicted TonB-dependent transmembrane protein (HpuB), homologous to siderophore receptors, and a lipoprotein (HpuA), which may be a homologue of TbpB/LbpB (Chen et al., 1996; Lewis et al., 1997). Purification experiments suggested that HpuA and HpuB proteins form a complex (Chen et al., 1998), and genetic data indicate that both are required for Hb utilization (Lewis et al., 1998). The HmbR and HpuB/HpuA receptors may remove haem from Hb by inducing a similar conformational change in Hb. HpuA plays a key role in this process but is also essential for binding and using Hb-Hp as an iron source for growth. Thus, HmbR might represent an ancestral form of haem/Hb receptor, with the lipoprotein component acquired later for increasing the efficiency of the process and extending the substrate profile.
Neisserial mutants in HmbR, HpuAB and TonB, although unable to use Hb and Hb-Hp, are capable of growing on haem as an iron source (Chen et al., 1996; Stojiljkovic et al., 1996; Stojiljkovic and Srinivasan, 1997). Thus, haem may diffuse passively across the outer membrane, as a result of its hydrophobic nature, or possibly through one or more porins. The TonB-independent haem uptake is probably of limited value in obtaining iron in vivo, as large amounts of free haem are rarely found on mucosal surfaces.
Preliminary biochemical studies have suggested that iron may be removed from haem at the cell surface and transferred to FbpA (Desai et al., 1995). However, the observation that mutants in FbpA can use haem as an iron source (Khun et al., 1998) indicates that the haem iron uses a different pathway. By analogy with the siderophore and Tf/Lf uptake systems, it would seem logical that a homologue of the Fbp ABC pathway might be responsible for mediating the transport of haem from the periplasm to the cytoplasm, similar to that found in Yersinia (Stojiljkovic and Hantke, 1994). Although a similar system has not been demonstrated in Neisseria, there are many putative binding proteins from the genome sequencing project that could serve as the haem-binding protein. Haem in the cytoplasm would be degraded rapidly by the action of an as yet uncharacterized haem oxygenase-like enzyme, thus providing iron to bacteria. Alternatively, haem may be incorporated directly into cytochromes and other haem proteins, thereby allowing cellular iron stores to be used for other purposes.
The receptor proteins involved in iron acquisition pathways are potentially ideal vaccine candidates because of their inherent surface accessibility and their role in survival and disease causation. This rationale has led to extensive studies evaluating the vaccine potential of the transferrin receptor proteins, particularly TbpB (Lissolo et al., 1995), which have recently been tested in phase I clinical trials (Danve et al., 1998). Receptor proteins from other iron acquisition pathways could serve as alternative vaccine candidates but, unless they provide specific advantages over TbpB, they may not receive serious consideration for vaccine development.
Transferrin receptors are present in all clinical isolates, have been shown to be important for infection in vivo (Cornelissen et al., 1998) and do not manifest phase variation. Although receptors for lactoferrrin, siderophores and haem iron sources may also be important for disease causation, their prevalence and expression properties are unlikely to serve as advantages over transferrin receptor proteins as vaccine candidates. However, there is considerable genetic and antigenic variation among TbpBs, such that at least two representative TbpBs will be required to provide broad-spectrum coverage (Rokbi et al., 1997). Further studies into the efficacy and antigenic variation of the other surface receptors will be required to determine whether they could provide advantages over a TbpB-based vaccine. Alternatively, the conserved binding interaction of TbpBs with Tf (Retzer et al., 1999) may ultimately be exploited to develop broad-spectrum vaccines.