Emerging strategies in microbial haem capture




Gram-negative pathogenic bacteria have evolved novel strategies to obtain iron from host haem-sequestering proteins. These include the production of specific outer membrane receptors that bind directly to host haem-sequestering proteins, secreted haem-binding proteins (haemophores) that bind haem/haemoglobin/haemopexin and deliver the complex to a bacterial cell surface receptor and bacterial proteases that degrade haem-sequestering proteins. Once removed from haem-sequestering proteins, haem may be transported via the bacterial outer membrane receptor into the cell. Recent studies have begun to define the steps by which haem is removed from bacterial haem proteins and transported into the cell. This review describes recent work on the discovery and characterization of these systems. Reference is also made to the transport of haem in serum (via haemoglobin, haemoglobin/haptoglobin, haemopexin, albumin and lipoproteins) and to mechanisms of iron removal from the haem itself (probably via a haem oxygenase pathway in which the protoporphyrin ring is degraded). Haem protein–receptor interactions are discussed in terms of the criteria that govern protein–protein interactions in general, and connections between haem transport and the emerging field of metal transport via metallochaperones are outlined.


Owing to their abundance in the host, haem-containing proteins are a potentially valuable source of iron for invading microorganisms. Intracellular pathogens can use haem directly; however, utilization of haem by extracellular pathogenic bacteria requires that haem-containing proteins first be released from the cell (Wandersman and Stojiljkovic, 2000). Haem release typically occurs by some form of tissue damage, resulting in the release of intracellular material. Once haem is released, it is bound by plasma proteins including haemopexin, albumin and lipoproteins (Camejo et al., 1998; Evans et al., 1999). To survive in the haem-limited environment of the host, microorganisms have developed diverse and elaborate systems to obtain this essential nutrient. The aim of this review is to discuss these systems in detail with particular emphasis on how these specific microbial haem capture systems function at the molecular level.

Haem availability in vivo

Formally, the term haem refers to reduced, ferrous or Fe(II) iron protoporphyrin IX. The term haemin refers to the oxidized, ferric or Fe(III) form of the molecule. In aqueous solution in the absence of proteins or reducing agents, iron protoporphyrin is found in its oxidized form (haemin). Although the formal definitions distinguish ‘haem’ and ‘haemin’, the term ‘haem’ is widely used to indicate iron protoporphyrin IX in either oxidation state. Haemin is usually considered to have either chloride or hydroxide (water) as an axial ligand. Haemin hydroxide converts readily to the µ-oxo form, a dimer with an Fe-O-Fe bridge, at high pH (Fig. 1). The hydroxide and µ-oxo forms of haemin intercovert readily. Dissociation of µ-oxo dimers (or larger aggregates) has a rate constant of 3–4 × 10−3 s−1; this can be rate limiting for some haemin–binding protein interactions (Hargrove et al., 1996; Kuzelovået al., 1997).

Figure 1.

Structure of haemin and haemoglobin.

Top. Structure of haemin and its µ-oxodimer.

Bottom. Structure of human haemoglobin, showing the position of the haems (left) and their exposures (right).

In serum, haem is bound primarily to haemoglobin, haemopexin, albumin and lipoproteins. Haemoglobin is the source of most circulating haem (Evans et al., 1999). It is found in the deoxyhaemoglobin [Fe(II), deoxyHb, no sixth ligand to the iron], oxyhaemoglobin [Fe(II), Hb, O2 as the sixth ligand to the iron] and methaemoglobin [Fe(III), metHb, no oxygen binding] forms. These three forms of the protein have somewhat different three-dimensional structures (Perutz et al., 1998). It is possible that they interact differently with bacterial haemoglobin receptors, providing redox/oxygen tension-associated control of haem uptake. Haemoglobin in serum is tightly bound to haptoglobin (a heterotetrameric glycoprotein) with a dissociation constant greater than 10−15 M (Evans et al., 1999). The serum concentration of haptoglobin in adults, ≈ 1.2 mg ml−1 (Evans et al., 1999), is sufficient to bind free haemoglobin.

Human albumin is a globular protein of ≈ 68 kDa (Peters, 1996). Human adult serum contains ≈ 40 mg ml−1 albumin or about 60% of the total protein content in serum (Dockal et al., 1999). The crystal structure shows that the protein has three major domains (Fig. 2). Haem is thought to bind most tightly to the first of the domains (Dockal et al., 1999), with a binding constant of ≈ 108 M−1 (Table 1). Although the first domain presumably has the highest binding affinity, all three domains bind haem (Dockal et al., 1999). It has been reported that HSA can bind up to 10 mol of haem mol−1 protein (Pasternack et al., 1985). HSA has been reported to bind haem ≈ 1000-fold more tightly than BSA (Benesch, 1994). However, the relative binding might depend on other molecules in solution.

Figure 2.

Structure of albumin and haemopexin.

Left. Structure of human albumin. The major binding site of haemin is at the upper right hand of the structure as drawn (Dockal et al., 1999).

Middle and right. Structure of haemopexin, showing the position of the haem and its two axial histidines (middle) and the haem exposure (right). The X-ray co-ordinates are from Paoli et al. (1999).

Table 1. Equilibria for binding of haem and haemin to proteins.


Iron porphyrin
Equilibrium constant
  1. a . Hargrove et al. (1994).

  2. b . Hrkal et al. (1974).

  3. c . Binding model assumes one primary binding site; Adams and Berman (1980).

  4. d . Binding model assumes two primary binding sites; Gattoni et al. (1996).

  5. e . Binding model assumes a dissociation constant (K d) and a Hill coefficient; the table gives 1/Kd.

  6. f . Hill coefficient of 1.68–2.52; Camejo et al. (1998).

  7. g . Benesch (1994).

  8. h . Beaven et al. (1974).

  9. i . Hill coefficient of 0.91–1.31 for LDL and 0.70–1.10 for HDL. See also Miller and Shaklai (1999).

Haemoglobin AHaem [Fe(II)]a> 1016
Haemoglobin AHaemin [Fe(III)]b> 1013
Human serum albuminHaeminc1.1 × 108
Haemind2.3 × 108
7.8 × 107
 Haemine,f1 × 106
Bovine serum albuminHaeming0.2 × 105
HaemopexinHaeminb,h1 × 1013
LDLHaemine,i2.0 × 107
HDLHaemine,i3.4 × 106

Haemopexin is a 63 kDa single polypeptide chain glycoprotein with multiple glycosylation sites (Muller-Eberhard, 1988). In humans, the serum haemopexin level is ≈ 10−5 M. Haem binds to haemopexin in an equimolar complex with a Ka of about 1013 M−1 (Table 1) (Evans et al., 1999). The affinity of haemopexin for haemin is higher than that of haemoglobin and several orders of magnitude higher than that of albumin (Evans et al., 1999). The haem is bound to two histidine residues (Fig. 2) (Paoli et al., 1999). Haem binding and release induce a substantial conformational change in the apoprotein. In line with this, recent studies have demonstrated that haem-loaded haemopexin differs from the apoprotein in its susceptibility to degradation by bacterial proteases (A. Sroka, J. Protempa, J. Travis, and C. A. Genco, unpublished).

Haem is also known to bind to lipoproteins. Camejo et al. (1998) have shown that haemin is about evenly distributed between lipoproteins and non-lipoprotein proteins. In a second study, as much as 80% of the total haemin was found to bind initially to low-density lipoproteins (LDL) and high-density lipoproteins (HDL) (Miller and Shaklai, 1999). This was followed by partial transfer of haemin to albumin and haemopexin. Both these studies noted that there were discrepancies between equilibrium constants measured in solutions of pure targets and the distribution of haemin among proteins in serum; this has yet to be resolved.

Microbial mechanisms for haem capture

Receptor-mediated uptake

In Gram-negative bacteria, the outer membrane forms a permeability barrier for substrates of > 600 Da (Braun and Killman, 1999). Haem itself has a molecular weight of just over 600. In solution, haem is a mixture of monomers and dimers (the monomer with a chloride or hydroxide ligand and the µ-oxo dimer with an Fe-O-Fe linkage). The size of haem in solution might limit its ability to transverse the bacterial outer membrane readily through porin channels. It is of interest to note, however, that haem can diffuse passively across model lipid bilayers. Transfer of haemin from liposomes is characterized by two rate constants; the slower rate has been ascribed to transmembrane movement of the haemin from one bilayer to the other with a half-time of seconds to ≈ 1 min. This flip-flop transmembrane movement is a function of the chain length of the lipid (Light and Olson, 1990) and, perhaps by analogy with porphyrins (Maman and Brault, 1998), of the protonation state of the propionates of the haemin.

The best-characterized mechanism by which bacteria acquire haem involves direct binding of haem or haem proteins to specific outer membrane receptors (Table 2 and Fig. 3). After the binding of haem or host haem-sequestering proteins to specific receptors, haem is removed from the bacterial receptor and transported into the cell by an energy-requiring process. Energy for the transport of iron or haem across the outer membrane in most Gram-negative organisms is provided by TonB in association with the ExbB and ExbD proteins (Braun, 1995). The TonB system uses the proton motive force of the cytoplasmic membrane for the passage of ligands into the periplasm. Receptors, which require energy supplied via the TonB system, are termed ‘TonB dependent’ and share amino acid homology in several regions termed ‘TonB boxes’. The TonB box represents the domain of the bacterial receptor that interacts physically with the energy-transducing protein TonB. In most Gram-negative pathogens that have been examined, a functional TonB protein is required for haem and haemoglobin utilization (Braun, 1995). Several Gram-negative organisms have a second TonB or TonB-like system that functions to transduce energy for iron accumulation from host iron-binding proteins (Occhino et al., 1998; O'Malley et al., 1999; Desai et al., 2000; Zhao and Poole, 2000).

Table 2. Microbial receptors for haem and haem-sequestering proteins.
MicroorganismProteinGeneSubstrateIdentification methodaReferenceb
  • a

    . Functions were defined genetically through the use of defined mutants and/or biochemically by binding to haem or host haem-sequestering proteins.

  • b

    . The most recent reference has been listed.

  • c

    . NI, not identified.

  • d

    . These proteins also function as soluble Hm- and Hb-binding proteins.

  • e

    . This is believed to be a low-affinity Hm receptor.

  • f

    . A mutant has not been constructed.

B. fragilis 44 kDaNIcHmBiochemical Otto et al. (1994)
C. diphtheriae HmuT hmuTHm and HbGenetic/biochemical Drazek et al. (2000)
E. coli ChuA chuAHm and HbGenetic/biochemical Torres and Payne (1997)
H. influenzae HgpA hgpAHb/HpGenetic/biochemical Morton et al. (1999)
HgpB hgpB Hb/HpGenetic/biochemical Morton et al. (1999)
HgpC hgpC Hb/HpGenetic/biochemical Morton et al. (1999)
HxuA hxuA Hm and HpxdGenetic/biochemical Cope et al. (1998)
P4 hel HmGenetic Reidl and Mekalanos (1996)
57 kDaNIHpxBiochemical Wong et al. (1995)
H. influenzae non-typeableHhuA hhuAHb/HpGenetic/biochemical Maciver et al. (1996)
HgbA hgbA Hb/HpGenetic/biochemical Cope et al. (2000)
HgbB hgbB Hb/HpGenetic/biochemical Cope et al. (2000)
HgbC hgbC Hb/HpGenetic/biochemical Cope et al. (2000)
H. ducreyi HgbA hgbA HbGenetic/biochemical Elkins et al. (1995)
TdhA tdhA HmGenetic Thomas et al. (1998)
H. pylori 77 kDaNIHmBiochemical Worst et al. (1995)
50 kDaNIHmBiochemical Worst et al. (1995)
48 kDaNIHmBiochemical Worst et al. (1995)
N. gonorrhoeae HpuAB hpuAB Hb/HpGenetic/biochemical Chen et al. (1998)
97 kDaNIHm and HbBiochemical Lee and Levesque (1997)
N. meningitidis HpuAB hpuAB Hb/HpGenetic/biochemical Lewis et al. (1998);
HmbR hmbR HbGenetic/biochemical Stojiljkovic et al. (1996)
P. gingivalis HmuR hmuR Hm and HbGenetic/biochemical Simpson et al. (2000)
Kgp kgp Hm and HbdGenetic/biochemical Simpson et al. (1999);
DeCarlo et al. (1999)
Tla tla HmeGenetic Aduse-Opoku et al. (1997)
HemR hemR UnknownGeneticf Karunakaran et al. (1997)
26 kDaNIHmBiochemical Bramanti and Holt (1993)
P. aeruginosa PhuR phuR Hm and HbGenetic Ochsner et al. (2000)
HasR hasR Hm and HbGenetic Ochsner et al. (2000)
S. dysenteriae ShuA shuA HmGenetic/biochemical Wyckoff et al. (1998);
Mills and Payne (1997)
S. marcescens HasR hasR Hm and HbGenetic/biochemical Letoffe et al. (1999)
V. alginolyticus NI hutA HmGenetic O'Malley et al. (1999)
V. cholerae HutA hutA HmGenetic/biochemical Occhino et al. (1998)
V. parahaemolyticus NI hutA HmGenetic O'Malley et al. (1999)
V. vulnificus HupA hupA Hm and HbGenetic/biochemical Litwin and Byrne (1998)
Y. pestis HmuR hmuR Hm and HmGenetic Thompson et al. (1999)
Y. enterocolitica HemR hemR Hm and HmGenetic/biochemical Stojiljkovic and Hantke (1994)
Figure 3.

Microbial mechanisms for haem capture. Red blood cells (RBCs) are degraded by a bacterial haemolysin, resulting in the liberation of haemoglobin (Hb) and haem (Hm). Although not depicted, it is well accepted that, once released, haem is bound to albumin and haemopexin. Haem may be transported into the bacterial cell by several mechanisms.

A. Direct binding of haemoglobin to a specific TonB-dependent outer membrane receptor; Hb and Hm are proposed to bind to discrete sites on the receptor.

B. Capture of haemoglobin and haemopexin by secreted proteins (haemophores), which deliver the haemoglobin or haemopexin to a specific TonB-dependent outer membrane receptor; the Hb receptor may bind the Hb-bound haemophore through either the Hb or the haemophore.

C. Degradation of haemoglobin, haemopexin and human serum albumin by bacterial proteases, resulting in the liberation of haem. Bacterial proteases can be either membrane bound or secreted. Haem is subsequently bound by TonB-dependent haem/haemoglobin receptors and transported into the periplasm. Energy for the transport of iron or haem across the outer membrane into the periplasmic space is provided by TonB in association with the ExbB and ExbD proteins. The transport process through the cytoplasmic membrane is not well described but appears to occur via an ABC transport system composed of a cytoplasmic membrane-associated permease and an ATPase. Within the cytoplasm, haem is degraded by a haem oxygenase.

Haem-specific receptors

Although numerous bacterial haem receptors have been identified, the binding constants of haem for these haem receptors and the specificity of the interaction have not been well defined. The majority of studies in which bacterial haem receptors have been defined have used genetic methods; only in a few cases has biochemical analysis been reported. In Vibrio cholerae, genetic and biochemical studies have confirmed the role of the outer membrane haem transport protein HutA in both haem binding and utilization (Occhino et al., 1998). HutA functions together with the proteins encoded by the hutBCD locus for transport of haem into the cell. The hutBCD locus is part of an operon that also includes tonB1, exbB1 and exbD1 genes, which function to provide energy for the transport of haem into the cell.

The haemin receptor HemR of Yersinia enterocolitica is encoded in a putative operon together with the hemSTUV genes (Wandersman and Stojiljkovic, 2000). Mutational analysis has defined two conserved histidine residues within HemR, His-128 and His-461, which appear to function in haem binding (Wandersman and Stojiljkovic, 2000). In Shigella dysenteriae, ShuA has been demonstrated to bind haem by affinity chromatography. The shuA gene is encoded within a discrete haem transport locus containing seven open reading frames (ORFs) including the shuU and shuV genes that are functionally equivalent to the Y. enterocolitica hemU and hemV genes (Wyckoff et al., 1998). A locus (hmuTUV) required for the acquisition of iron from haemin and haemoglobin has been described recently in the Gram-positive organism Corynebacterium diphtheriae (Drazek et al., 2000). The hmuT gene encodes a lipoprotein that binds to both haemin and haemoglobin, and the authors of this study have postulated that the HmuT protein functions as a haem receptor in this organism.

Two putative haem receptors (Tla and HemR) have been described in Porphyromonas gingivalis, but a definitive role for Tla or HemR in haem binding has not been established. In the case of Tla, genetic data support a role for Tla in haem utilization (Aduse-Opoku et al., 1997). Conclusive genetic or biochemical evidence for the role of HemR in haem binding or utilization has not been obtained (Karunakaran et al., 1997). A P. gingivalis 26 kDa outer membrane protein that binds haem has also been described (Bramanti and Holt, 1993); however, the gene encoding this putative receptor has not been identified.

Yersinia pestis contains at least two distinct systems for haem utilization. The first system consists of the genes, hmuRSTU (Thompson et al., 1999). A second system that is homologous to the Serratia marcescens hasA system (encodes an extracellular haem-binding protein) has been identified in the complete Y. pestis genome sequence (Thompson et al., 1999). An independent locus in Y. pestis is involved in haemin storage (hmsFRS). The haem storage locus of Y. pestis is required for the transmission of plague from fleas to mammals (Hinnebusch et al., 1996) but does not appear to be required for growth of Y. pestis in mammals (Lillard et al., 1999).

Two distinct haem uptake systems have been described in Pseudomonas aeruginosa (Ochsner et al., 2000). The phu locus consists of the phuR receptor gene together with the phuSTUVW operon that encodes a typical ABC transporter. The second haem uptake system in P. aeruginosa is composed of genes homologous to the S. marcescens hasR and hasA genes. In Haemophilus influenzae, the lipoprotein e (P4) is essential for the acquisition of haem, but the ability of lipoprotein e to bind haem has not been reported. Reilly et al. (1999) have shown recently that this outer membrane lipoprotein is a phosphomonoesterase, and they suggested that this activity could result in haem modification or alteration of the phosphorylation states of other components in a putative haem transport complex.

Haemopexin receptors

The only bacterial haem transport protein known to bind haemopexin directly is the H. influenzae HxuA protein, which functions at both the outer membrane level and as a soluble haem-binding protein (Cope et al., 1998). The amino acid sequence of HxuA shows no similarity to those of TonB-dependent outer membrane proteins, suggesting that HxuA might function as a haemophore rather than as a classical haem outer membrane receptor (see below). Three additional H. influenzae proteins have also been described that bind haem–haemopexin in affinity chromatography experiments (Wong et al., 1995). Genetic studies confirming the role of these proteins in haem–haemopexin uptake by H. influenzae have not been reported.

Haemoglobin receptors

Two haemoglobin receptors have been described in Neisseria meningitidis, HmbR and HpuB. Genetic studies have confirmed that HpuB and the lipoprotein HpuA are required for the utilization of either haemoglobin or haemoglobin–haptoglobin complexes (Lewis et al., 1998). Results obtained in a dot-blot assay using biotinylated haemoglobin suggest that HpuB binds the haemoglobin–haptoglobin complex via haptoglobin; however, definitive biochemical evidence for the specificity of this binding has not been reported. The role of hmbR in haemoglobin utilization has been confirmed in N. meningitidis by genetic analysis, and the specificity of binding of HmbR to haemoglobin has also been examined by dot-blot analysis (Stojiljkovic et al., 1996). HmbR does not appear to be functional in Neisseria gonorrhoeae; rather N. gonorrhoeae uses haemoglobin through the HpuB/HpuA system (Chen et al., 1998). Haem transport in N. meningitidis and N. gonorrhoeae occurs independently of HmbR and HpuB/HpuA; however, a specific haem receptor has not been identified.

Three haemoglobin/haemoglobin–haptoglobin receptors have been identified in H. influenzae, HgpA, HgpB and HgpC; any one of these proteins is sufficient to support growth with haemoglobin–haptoglobin as the haem source (Morton et al., 1999). Disruption of hgpA, hgpB and hgpC still enables H. influenzae to grow with haemoglobin, indicating that an additional haemoglobin acquisition mechanism exists. Biochemical and genetic studies have confirmed the role of the Haemophilus ducreyi haemoglobin receptor HgbA in haemoglobin utilization in H. ducreyi (Elkins et al., 1995). An isogenic hgbA mutant has been shown to be highly attenuated in the human challenge model of H. ducreyi infection (Al-Tawfig et al., 2000).

Genetic studies have confirmed that the P. gingivalis TonB-dependent haemoglobin receptor (HmuR) is required for the acquisition of both haem and haemoglobin (Simpson et al., 2000). HmuR has also been demonstrated to bind both haemoglobin and haem by standard biochemical binding assays (T. Olczak, D. W. Dixon, and C. A. Genco, unpublished). Genetic evidence indicates that the lysine-specific cysteine proteinase gingipain K (Kgp) is also required for the utilization of haemoglobin by P. gingivalis (Simpson et al., 1999). Furthermore, biochemical evidence supports a role for Kgp in both haemin and haemoglobin binding (DeCarlo et al., 1999; Shi et al., 1999; T. Olczak, D. W. Dixon, and C. A. Genco, unpublished; A. Sroka, J. Protempa, J. Travis, and C. A. Genco, unpublished). Like the H. influenzae HxuA protein, Kgp can be found both associated with the outer membrane and in a soluble form. It has been proposed that Kgp functions primarily in its soluble form to bind haem and haemoglobin and to deliver haem to the outer membrane receptor HmuR (T. Olzcak, D. W. Dixon, and C. A. Genco, unpublished).

Haemophore-mediated uptake

Several Gram-negative bacteria produce extracellular haem/haemoglobin/haemopexin-binding proteins that capture and shuttle haem to a specific outer membrane receptor (Fig. 3). These secreted proteins function to extract haem from haemoglobin or haemopexin and deliver the haem complex to an outer membrane-associated protein that transports haem into the cell. The best characterized of these systems is that of the S. marcescens secreted protein, HasA, which extracts haem from haem proteins and delivers it to the outer membrane receptor HasR (Létofféet al., 1999). A gene encoding a similar extracellular haemin-binding protein (HasAp) has also been described in P. aeruginosa and Pseudomonas fluorescens (Létofféet al., 1998; 2000). HasR alone can transport free haem or haem from haemoglobin, but the presence of HasA greatly facilitates haem uptake. HasA binds one haem per molecule with high affinity, and binding does not appear to modify the conformation of HasA. For equimolar HasA and haemoglobin, the haem is mainly in the HasA protein, implying that HasA has the higher affinity. Haem transfer does not appear to involve a complex between HasA and haemoglobin; therefore, Létofféet al. (1999) have proposed that the mechanism of haem transfer involves initial loss of the haem from haemoglobin into solution followed by uptake by the HasA protein. Direct transfer is also a possibility, however, and would involve a transient complex. This process might circumvent the tendency of free haemin to form the µ-oxo dimer.

The P. gingivalis lysine-specific cysteine protease Kgp also functions as a bacterial haemophore, binding both haemin and haemoglobin (DeCarlo et al., 1999; T. Olzcak, D. W. Dixon, and C. A. Genco, unpublished; A. Sroka, J. Protempa, J. Travis, and C. A. Genco, unpublished). The P. gingivalis TonB-dependent haemoglobin receptor HmuR has been shown to bind soluble Kgp (T. Olczak, D. W. Dixon, and C. A. Genco, unpublished), and it has been proposed that Kgp and HmuR function together for the transport of haemin from haemoglobin, via a haemophore-like system. In a human pathogenic strain of Escherichia coli (EB1), the extracellular Hbp protein has been shown to bind haem and haemoglobin (Otto et al., 1998). The secreted H. influenzae haem/haemopexin-binding protein HxuA might function in a manner similar to S. marcescens HasA, P. gingivalis Kgp and E. coli Hbp. Like the P. gingivalis Kgp protein, HxuA is both cell surface associated and secreted into the culture supernatant. The protein product of the H. influenzae hxuB gene has been proposed to function as the receptor to which soluble HxuA binds (Cope et al., 1998).

Haemoglobin proteases

An additional mechanism for the acquisition of haem from host haem-sequestering proteins involves an extracellular protease that degrades host haem proteins (Fig. 3). Biochemical and genetic studies have confirmed that the P. gingivalis lysine-specific cysteine proteinase Kgp can both bind and degrade soluble haemoglobin as well as haemopexin, haptoglobin and transferrin in normal human serum (DeCarlo et al., 1999; Lewis et al., 1999; A. Sroka, J. Protempa, J. Travis, and C. A. Genco, unpublished). Kgp is highly efficient at haemoglobin degradation, complete digestion being observed at nmol concentrations of enzyme within 2–3 h (A. Sroka, J. Protempa, J. Travis, and C. A. Genco, unpublished).

The E. coli EB1 haemin-binding protein Hbp can also degrade haemoglobin. In an initial report, Hbp was shown to degrade ≈ 300 pmol of haemoglobin mg−1 protein within 4 h (Otto et al., 1998). However, it is not known whether the E. coli haemoglobinase can cleave haemoglobin in serum (where inhibitors could restrict proteolytic activity, and where haemoglobin is complexed to haptoglobin).

Energy requirements for loss of haem from host haem-sequestering proteins

For microbial haem capture and utilization from host serum haem-sequestering proteins, haem must ultimately be removed from the protein. Whether energy is necessary to induce loss of the haem from the host haem-sequestering protein probably depends largely on the affinity of the bacterial haem-binding protein for the haem. Thus, proteins with moderate affinity (Ka < 109 M−1) are likely to have off rates that are on a useful timescale for growth via iron derived from haem, k(off) > 0.1 s−1[Ka = k(on)/k(off)]. Proteins that bind the haem far more tightly might need to undergo an energy-induced conformational change to lose haem on a useful timescale. Thus, a protein that binds haem with a Ka = 1013 M−1, would have an off rate of 10−5 s−1 or a time constant for haem loss of approximately 1 day. Addition of energy to the system can substantially increase the rate of haem loss. Approximately half the energy from a single ATP (i.e. half of 7.3 kcal mol−1) could increase the rate constant more than 1000-fold. Thus, unless the haem has an unusually slow off rate because it is very tightly bound (high Ka) or deeply buried (moderate Ka, but slow on and off rates), haem transfer from the host haem-sequestering protein to the bacterial haem transport protein is expected to require no more than modest energy.

Haem interactions with bacterial outer membrane receptors and soluble haem-binding proteins

In thinking of recognition of host haem-binding proteins by bacterial proteins that will remove or transport the haem, it is useful to consider the possible parameters of the recognition site between the proteins. Haem has an accessible surface area of ≈ 800 Å2 (Stellwagen, 1978). For haemoglobin, the percentage of the haem exposed to solvent is about 15% (123 Å2) for the α-subunits and 20% (173 Å2) for the β-subunits (Stellwagen, 1978). For haemoglobin bound to a partner protein, haem can certainly serve as part of the recognition site. However, it is unlikely to serve as the only point of recognition. A recent analysis of 75 two-component protein–protein complexes found no protein–protein complex with a recognition site smaller than 1000 Å2 (Lo Conte et al., 1999). With regard to recognition of the haem in haemoglobin by another protein, a minimum protein–protein recognition site of 1000 Å2 implies that the haem itself will form only a minor portion of the recognition site, probably less than 20%. The rest of the recognition will result from the haemoglobin protein itself.

In general, the receptor domains involved in binding of haem have not been well characterized in any of the known bacterial haem receptors. For bacterial receptors that recognize both haem and haemoglobin, it is not known whether there are discrete binding sites for haem and haemoglobin or more than one site that recognizes a common motif. In general, haem proteins use a variety of axial ligands including histidine, tyrosine and cysteine single ligands and bishistidine, histidine–methionine, bismethionine, histidine–tyrosine and histidine and the N-terminal amine ligand pairs (Poulos, 1996). Amino acid comparisons of the conserved domains of several bacterial haem and haemoglobin receptors has revealed a conserved receptor domain containing invariant histidine residues, FRAP and NPNL amino acid boxes. However, only in the case of the S. marcescens HasA and Y. enterocolitica HemR proteins have specific axial ligands been defined (Létofféet al., 1999; Wandersman and Stojiljkovic, 2000).

Haem removal from bacterial outer membrane receptors and soluble haem-binding proteins

In the case of bacterial receptors that bind haem proteins, it is generally believed that haem is removed from the protein and only haem or iron is transported into the cell. Theoretically, once removed from host haem-sequestering proteins, haem can be transported via the bacterial haem receptor, or the Fe from haem may be removed and iron transported across the outer membrane through a separate iron-specific transport process. For the bacterial extracellular haem-binding proteins HasA, Kgp and HxuA, the protein must first capture haem from haemoglobin, but must then release this haem onto the outer membrane receptor. How this transfer occurs is not known.

Haem transport proteins would be expected to have similar requirements to other metallochaperones (O'Halloran and Culotta, 2000). Haem transport proteins must obtain haem either from free haem in solution or from another protein, transport it to the target and then dock to the target in order to transfer the haem. The transfer process itself need not result in the release of free haem, as has been proposed for the S. marcescens HasA to HasR transfer (Létofféet al., 1999). Release is presumably favoured by some aspect of the interaction of the transport protein with its target, resulting in a conformational change in the transport protein. This could occur via structural changes induced by electrostatic or van der Waals' interactions, pKa changes of protein residues, phosphorylation of appropriate residues on either of the partner proteins or, for the haem itself, changes in the oxidation state or ligand binding. Once haem is removed from the bacterial outer membrane receptor, the haem is transported into the cell. This step is believed to occur via a TonB-induced conformational change in the bacterial outer membrane receptor, although this has not been demonstrated experimentally.

Transport of haem into the bacterial cell

Organisms such as P. gingivalis that have an absolute requirement for PPIX because they are deficient in one or more steps in the synthesis of this cofactor transport the entire haem molecule into the cell (Genco et al., 1994). Genetic studies have also demonstrated that organisms that do not have a strict requirement for PPIX also transport haem into the cell (Mills and Payne, 1995). From a strictly energy standpoint, transport of haem into the bacterial cell requires less energy then the biosynthesis of the PPIX ring. In some instances, the mechanism used for transport of iron into the cell could depend on the growth conditions. For example, biochemical studies have suggested that N. gonorrhoeae does not transport the entire haem into the cell (Desai et al., 1995). However, characterization of Neisseria hemA mutants indicates that these mutants can grow with haem as a porphyrin source (Lewis et al., 1998). Thus, both mechanisms for iron utilization from haem could operate under different growth conditions.

Loss of iron from haem might occur at the outer membrane, in the periplasm or in the cytoplasm. Mechanistically, haem-bound iron could be decomplexed by either (i) enzymatic ligand destruction with concomitant release of iron (haem oxygenase mechanism); or (ii) a simple enzymatic iron removal (reverse ferrochelatase mechanism). The first of these, a haem oxygenase mechanism, is discussed in more detail below. A reverse ferrochelatase has not yet been characterized either genetically or biochemically. An early report found intracellular fluorescence in H. influenzae attributed to PPIX that was taken to be suggestive of a reverse ferrochelatase (Loeb, 1995); however, there is no definitive evidence for the existence of such an enzyme.

Transport of haem or iron across the cytoplasmic membrane is driven by ATP hydrolysis. Bacterial ABC transport systems are commonly used for transport through the cytoplasmic membrane (Fath and Kolter, 1993). Although genes encoding putative ABC transporters have been identified in loci adjacent to several haem and haemoglobin receptors, the role of the proteins encoded by these loci have not been defined. Likewise, the mechanisms of transport through the cytoplasmic membrane is not known.

Once internalized in the cytoplasm, haem must be degraded to release iron. Recent studies have shown that, at least for the Gram-positive organism C. diphtheriae, this occurs via a bacterial haem oxygenase (Chu et al., 1999). Haem oxygenases have been well characterized from eukaryotic systems and are involved in the oxidative degradation of haem through the cleavage of the porphyrin ring with production of CO, iron and biliverdin. In C. diphtheriae, the HmuO protein, unlike the membrane-bound mammalian haem oxygenase, is present in the bacterial cytosol. The bacterial HmuO has a considerably slower catalytic rate than the mammalian haem oxygenase (Chu et al., 1999). A recent study has identified a hmuO homologue in N. meningitidis, but conclusive evidence for the role of the product of this gene in haem degradation in Neisseria has not been obtained (Zhu et al., 2000).

Concluding remarks

Gram-negative pathogens are armed with an array of clever mechanisms for the capture of host haem. The development of these specialized systems for haem transport appears to be closely related to the specific nutritional requirements of the pathogen as well as the environmental niche in which the organism is found in vivo. We are now just beginning to understand the details of how these specific microbial haem capture systems function at the molecular level. Future studies should be directed at determining the specificity of bacterial haem transport proteins for haem, defining the steps by which haem is removed from bacterial haem-transporting proteins and unravelling the mechanisms by which haem is actually transported into the bacterial cell. It will also be important to determine whether specific bacterial receptors for haem bound to lipoproteins are present in Gram-negative bacteria.


C.A.G.'s studies are supported by Public Health Service Grant AI30797 from the National Institute of Allergy and Infectious Diseases and grant DE09161 from the National Institute of Dental and Craniofacial Research, and D.W.D.'s by grant AI45883 from the National Institute of Allergy and Infectious Diseases. We thank John Olson for helpful discussions, Jack Murphy for critical review of the manuscript, and Waltena Simpson for assistance with figures.