The periplasmic FbpA (ferric-binding protein A) from Haemophilus influenzae plays a critical role in acquiring iron from host transferrin, shuttling iron from the outer-membrane receptor complex to the inner-membrane transport complex responsible for transporting iron into the cytoplasm. In the present study, we report on the properties of a series of site-directed mutants of two adjacent tyrosine residues involved in iron co-ordination, and demonstrate that, in contrast with mutation of equivalent residues in the N-lobe of human transferrin, the mutant FbpAs retain significant iron-binding affinity regardless of the nature of the replacement amino acid. The Y195A and Y196A FbpAs are not only capable of binding iron, but are proficient in mediating periplasm-to-cytoplasm iron transport in a reconstituted FbpABC pathway in a specialized Escherichia coli reporter strain. This indicates that their inability to mediate iron acquisition from transferrin is due to their inability to compete for iron with receptor-bound transferrin. Wild-type iron-loaded FbpA could be crystalized in a closed or open state depending upon the crystallization conditions. The synergistic phosphate anion was not present in the iron-loaded open form, suggesting that initial anchoring of iron was mediated by the adjacent tyrosine residues and that alternate pathways for iron and anion binding and release may be considered. Collectively, these results demonstrate that the presence of a twin-tyrosine motif common to many periplasmic iron-binding proteins is critical for initially capturing the ferric ion released by the outer-membrane receptor complex.
Since iron is an essential element for many biological processes, there are a variety of specialized systems for transport, uptake and storage of this metal-ion cofactor in vertebrates that limit its availability to micro-organisms . The primary mode of iron transfer throughout the body is via the glycoprotein transferrin. The transfer of iron to transferrin from dietary sources or from recycling pathways in the host is facilitated by ceruloplasmin or hephaestin, but transferrin is also capable of readily scavenging free ferric ion that it encounters. The ability to scavenge free ferric ion is also a characteristic feature of the related glycoprotein lactoferrin, particularly at sites of inflammation and other areas of lower pH . Collectively, these proteins are responsible for sequestering free ferric ion in the host such that extracellular microbes require efficient iron-acquisition systems to survive and grow in the various ecological niches within the host.
Transferrin and lactoferrin are 80 kDa bi-lobed proteins, with each lobe being capable of binding a single iron atom. Each lobe is comprised of two structured domains that are connected by several β-strands with the site of iron binding at the base of the interdomain cleft. The β-strands facilitate substantial rigid-body conformational changes that range from a closed conformation with the domains in close juxtaposition to an open conformation involving rotational movements of up to 54 ° between the two domains . The apoprotein is preferentially in the open conformation and, upon binding of a ferric ion to the C-terminal domain, a subsequent rotation of the N-terminal domain completes the iron-co-ordination complex and stabilizes the iron-loaded form in the closed conformation .
The current model proposes that the process is initiated by binding of a synergistic carbonate anion to the C-terminal domain forming a cluster of four ligands that facilitate iron binding (two tyrosine residues and carbonate). The critical role of the synergistic anion is supported by biochemical studies with proteins altered in the anion-binding ligands . Structural studies with iron-loaded camel lactoferrin in the open conformation, obtained by a microdialysis method, directly demonstrated that the bound iron atom is co-ordinated by two tyrosine residues and a carbonate anion . Site-directed mutagenesis of the individual ligands involved in co-ordinating iron demonstrate that Tyr188 of human transferrin is essential for binding, whereas mutation of Tyr95  or the histidine or glutamic acid residues [7,8] reduce, but do not eliminate, iron binding. The implication is that Tyr188 is critical for the formation of the iron-co-ordination complex and may serve as the initial binding contact.
The release of iron from transferrin occurs in the endocytic vesicle, facilitated by the low pH. Protonation and release of the carbonate anion is proposed to be the first step in iron release, followed by protonation of the histidine ligand . A pH-sensitive di-lysine trigger has also been proposed to be involved in the release of iron from transferrin , but previous studies suggest that the primary effect of mutation is to indirectly weaken iron co-ordination and facilitate the release of the synergistic anion . In contrast with transferrin, lactoferrin retains iron at relatively low pH levels, consistent with its role in iron sequestration at sites of inflammation. The retention of iron at low pH has been attributed to co-operation between the two lobes in lactoferrin, which may be influenced in part by the α-helical connecting peptide between the lobes that is relatively non-structured in transferrin .
The periplasmic iron-binding proteins from Haemophilus influenzae and Neisseria meningitidis are considered ‘bacterial transferrins’ because of their structural similarity to a single lobe of transferrin or lactoferrin, consisting of two domains connected by a pair of antiparallel β-strands [12,13]. Analogous to transferrin, the ferric ion is co-ordinated by a similar set of ligands in an octahedral geometry: two tyrosine residues, a histidine residue, a glutamic acid residue, a phosphate anion and a water molecule. These Fbps (ferric-binding proteins) are essential for acquiring iron from host transferrin or lactoferrin [14,15] and are responsible for shuttling iron from an outer-membrane receptor complex, that binds to and removes iron from the host glycoprotein, to an inner-membrane transport complex that transports iron to the cytoplasm . The outer-membrane complex, composed of a surface lipoprotein, TbpB (transferrin-binding protein B), and an integral outer-membrane protein, TbpA (transferrin-binding protein A), may simply release iron into the periplasm where it is efficiently scavenged by FbpA. However, recent studies demonstrating a TbpA–FbpA interaction  suggest that the iron from transferrin may be transferred directly from TbpA to FbpA. The inability of FbpA mutants with reduced iron-binding affinities to mediate acquisition of iron from transferrin  suggests that high-affinity binding by FbpA is required to drive the transport process and implies a competition between FbpA and receptor-bound transferrin. Although there is currently little information available on the inner-membrane transport complex or mechanism of iron removal from FbpA and transport across the inner membrane, it is likely that these involve the ATP-driven conformational changes and transport cycle proposed for ABC (ATP-binding cassette) importers . However, it is not clear that domain separation alone would be sufficient for effective release of ferric ion bound to FbpA, thus a greater understanding of the iron-release process will be required.
In spite of different physiological roles, the similarities in structure and iron co-ordination between eukaryotic transferrins, lactoferrins and the bacterial transferrins  suggest that there may be common basic mechanisms for iron binding and release. Considerable insights into the mechanism of iron binding and release by transferrins and lactoferrins have been obtained through site-directed mutagenesis followed by structural and functional studies with the recombinant proteins . This approach has been used less extensively to probe the iron-binding process of FbpA and to date has not yielded structures of the complete iron-co-ordination complex of mutant proteins in the closed conformation [18,20–22]. One apparent difference between human transferrin and FbpA is the role of the tyrosine ligands, as mutation of a single tyrosine residue eliminates iron-binding capabilities of the human transferrin N-lobe , but not of FbpA . In the present study, a series of site-directed mutants of the conserved tyrosine ligands of FbpA were prepared to probe their role in iron binding under in vitro and in vivo conditions. In parallel, protein crystallography studies were pursued to obtain structures of iron-loaded proteins in the open and closed conformation to gain further insights into iron co-ordination.
Further details are provided in the Experimental section in the Supplementary material at http://www.BiochemJ.org/bj/432/bj4320057add.htm. Mutant H. influenzae fbpA genes were prepared by SOEing (splicing by overlap extension) PCR or the QuikChange™ mutagenesis protocol and subcloned into a vector used for protein production. The recombinant FbpA proteins were expressed at high levels in the Escherichia coli periplasm, and pure preparations of protein were obtained after ion-exchange chromatography for spectral analyses and crystallography. Spectra were obtained for iron-loaded preparations of FbpAs, and the iron-binding properties were assessed using a citrate-competition assay (Table 1). As described more fully in the Supplementary Experimental section, Supplementary Table S1 at http://www.BiochemJ.org/bj/432/bj4320057add.htm. and the Results section below, the purified FbpA preparations were also used in crystallization screens and structures of several wild-type and mutant FbpAs were determined. The four refined crystal structures of FbpA determined for the present study have been deposited in the PDB under the accession codes 3KN7 (Y195A), 3KN8 (Y196A), 3OD7 (wild-type closed) and 3ODB (wild-type open). For statistics, see Supplementary Table S2 at http://www.BiochemJ.org/bj/432/bj4320057add.htm.
|FbpA||Mass absorption coefficient||Molar absorption coefficient||λmax (nm)||Affinity constant (logKa)|
|FbpA||Mass absorption coefficient||Molar absorption coefficient||λmax (nm)||Affinity constant (logKa)|
Properties of the site-directed mutant FbpAs
The two tyrosine ligands (Tyr195 and Tyr196) of the ‘bacterial transferrin’, FbpA from H. influenzae, were targeted for site-directed mutagenesis in order to evaluate their contribution to iron co-ordination and compare it with that of vertebrate transferrins and lactoferrins. A set of site-directed mutants of Tyr195 and Tyr196 were designed to provide a range of amino acid replacements of varying size and properties (Table 1). The combination of high-level expression by the T7 expression system and efficient export into the periplasmic space provides a facile method for obtaining relatively pure preparations of FbpA. The level of production is usually in the 50–150 mg/l range and wild-type FbpA usually constitutes >80% of the protein in the osmotic-shock fluid as assessed by SDS/PAGE (results not shown). Although there was a 2–3-fold variation in the level of protein expression for the different FbpAs, this may not reflect differences in protein stability, but, instead, the inherent variation in expression with this system. The purity of the proteins after ion exchange was comparable and, as there was no indication of instability of the resulting preparations, it was unlikely to have an impact on the results.
As a first step for probing the iron-binding properties of the different FbpA proteins, the purified protein preparations were subjected to the iron-loading procedure and the UV-visible (300–800 nm) spectra were determined. After removing the excess iron, colour was observed in all the protein samples except the Y195A/Y196A double mutant and a peak in the spectra in the 450–480 nm range was evident for the other mutants (Table 1). The λmax is inversely proportional to the energy of the charge-transfer band for the iron–tyrosine residue interaction. The absorption bands for all the mutants with the exception of Y195E have a considerable blue shift; such a shift suggests an increased Tyr(O)–Fe interaction with the remaining tyrosine residue. Since the Y195A/Y196A double mutant does not contain any iron-co-ordinating tyrosine residues, there are no detectable spectral changes, which is a useful control for the spectral measurements, but does not provide any information regarding iron binding.
In order to obtain an overall estimation of the iron-binding affinity of the mutant proteins, citrate-competition assays were performed. In this assay, 100 μM iron-loaded protein was exposed to varying concentrations of citrate (0.5–500 mM). This assay provides a relative measure of iron-binding affinity of the various mutants and the strength of the tyrosine residue–iron interaction. The concentration required to remove 50% of the iron for the mutant proteins is illustrated in Figure 1. The titration with citrate can be used to estimate the iron-binding constants for the mutant proteins, which ranged from 18.92 to 19.62 (Table 1) . It is salient to note that the present assay was performed at pH 8 to facilitate comparison with previous studies, but that the pH of the periplasm would normally be substantially lower and have an impact on the binding constants.
Mutations that replaced tyrosine residues with histidine or glutamic acid were included to provide the potential to co-ordinate iron, whereas isoleucine, phenylalanine and alanine residues would eliminate one iron-co-ordinating ligand. The potential for iron co-ordination clearly did not result in enhanced retention of iron as the mutant proteins with glutamic acid had the weakest iron-binding properties (Figure 1 and Table 1), and for each tyrosine residue, the alanine replacement tended to have the strongest binding. The overall trend for both Tyr195 and Tyr196 was Ala>His>Ile>Phe>Glu. The results suggest that the different side-chain replacements may be perturbing the iron-co-ordination geometry to varying degrees, indicating that structural studies will be required to explore the details of iron co-ordination in the various mutants.
Although the tyrosine mutants retain substantial iron-binding capabilities based on the in vitro binding studies (Figure 1 and Table 1), previous studies indicate that they are defective in supporting growth on transferrin or ferric citrate medium . The growth properties of H. influenzae expressing the Y195A or Y196A mutant proteins on iron-limited anaerobic growth medium were not distinguishable from a strain lacking an FbpA protein. This implies that there is either a deficiency in acquiring iron in the periplasm or a deficiency in donating iron to the inner-membrane complex composed of FbpB and FbpC. However, the inability to detect a difference between growth of strains expressing the mutant FbpAs and strains lacking FbpA does not necessarily imply that they are totally defective in iron transport. The presence of alternate iron pathways could readily obscure any enhancement that transport with the tyrosine mutant FbpAs would provide over an FbpABC pathway lacking FbpA. Thus an alternate-pathway-reconstitution approach was sought to address this issue.
An E. coli reporter strain, H1771, that is defective in iron uptake  has been used successfully to identify foreign ferric and ferrous ion-uptake pathways  and was selected for pathway-reconstitution experiments. This reporter strain is defective in a ferrous ion-uptake pathway (Feo-negative), resulting in reduced levels of intracellular iron, which is monitored by the presence of a β-galactosidase reporter gene fused to the Fur-regulated promoter region of the fhuF gene. There are substantial levels of β-galactosidase activity even when this strain is grown in medium containing high levels of exogenous iron, and thus repression of expression is dependent on the introduction of a functional iron-transport pathway. Preliminary experiments with plasmid pSC590 encoding the H. influenzae fbpABC operon  demonstrated the utility of this approach, as colonies of transformants with this plasmid were white on specialized MacConkey plates, compared with the red colony colour of the parent reporter strain. The presence of this vector resulted in substantial repression of β-galactosidase expression in cells grown on iron-supplemented medium (914–1345 units compared with cells carrying the control plasmid with 230915–254302 units). Thus it was decided to test the ability of this system to compare the function of mutant FbpAs (Figure 2).
Inhibition of β-galactosidase activities of strains expressing mutant FbpAs
The fbpA gene in the pSC590 vector was mutated using QuikChange™ mutagenesis as described in the Supplementary Experimental section. In order to account for any iron transport mediated by the FbpBC inner-membrane complex, a control plasmid was introduced in which two stop codons had been inserted in place of the tyrosine residues in the gene encoding FbpA (YYSTOP).
To evaluate the effect of mutation of the tyrosine residues, the pSC590 plasmid was mutagenized and transformed into strain H1771, producing the following derivatives: the Y195A mutant, the Y196A mutant and the Y195A/Y196A double mutant (YYAA). Three or more colonies were selected from several transformations for analysis in order to account for variability in the assay. The results illustrated in Figure 2 demonstrate that both of the Y195A and Y196A proteins are capable of acquiring iron in the E. coli periplasm and donating the iron to the inner-membrane transport complex. In contrast, the level of β-galactosidase activity in the strain expressing the double-mutant protein (YYAA) is comparable with the strain carrying the mutant gene containing stop codons (YYSTOP). This indicates that the Y195A/Y196A mutant FbpA is defective in iron binding or in donation to the inner-membrane transport complex, logically the former. The small residual amount of β-galactosidase activity in these cells relative to the empty vector (CV30) may be due to the presence of the functional FbpBC complex.
Although this assay was effective at addressing the questions posed, the results in Figure 2 illustrate that it may be limited in making comparisons between the different mutations. Owing to the experimental variation in the current assay, it is not possible to conclude that the Y195A FbpA is more effective at mediating iron acquisition than the Y196A FbpA. Thus it was not deemed appropriate for comparing the function of the various mutant FbpAs described in the previous section. It is evident that an alternate reporter system will need to be developed in order to be able to effectively compare these various mutant FbpAs.
Structural studies with site-directed mutants of human lactoferrin and human transferrin have provided considerable insights into iron co-ordination, as the mutant proteins crystallized in the closed conformation [7,8]. In contrast, structural studies with site-directed mutants of H. influenzae FbpA yielded structures of the mutant proteins in the open conformation [20–22], so information on how the mutation influenced the formation of the iron co-ordination complex is lacking. Therefore a strategy was pursued to use crystals of the wild-type protein in the closed conformation to seed formation of crystals of the mutant proteins. Crystallization setups using a common preparation of wild-type iron-loaded FbpA were prepared with an imidazole/malate buffer that was used in the first published study  and with a methyl ethane sulfonate buffer used in structural studies of the mutant proteins . Coloured crystals of the wild-type protein were readily obtained and used in a varietly of different seeding strategies with the mutant proteins. The wild-type crystals and mutant crystals were subjected to X-ray diffraction analysis and unit cell dimensions were determined as these could be used to discriminate between the closed and open conformations of FbpA (see the Supplementary Experimental section). Surprisingly, the only crystals in the closed conformation based on this analysis was the wild-type protein crystallized with the imidazole/malate buffer, suggesting that the wild-type protein crystallized in the methyl ethane sulfonate buffer was in the open conformation, yet contained bound iron.
Crystals of the wild-type protein in the closed conformation were prepared and used to seed the formation of crystals of mutant proteins, including the Q58L FbpA as this mutation had the least impact on iron binding . Although crystals were obtained in a variety of conditions with different mutant proteins, the crystals obtained under the seeding conditions had unit cell dimensions, indicating that the proteins were in the open conformation. The failure to obtain crystals of the mutant proteins in the closed conformation after repeated attempts at seeding suggested that it would be unlikely to obtain crystals of mutant proteins in the closed conformation, even after extensive screening, and prompted us to abandon these efforts.
The initial X-ray diffraction results suggesting that the wild-type iron-loaded FbpA crystallized in the open conformation were surprising since the iron-loaded forms of native transferrins, lactoferrins and periplasmic iron-binding proteins have crystallized in the closed conformation, except when specialized approaches were taken [5,27]. The same preparation of wild-type FbpA was used for the generation of crystals in the closed and open forms, indicating that crystallization in the open conformation was not due to modifications during the production and purification processes. Since comparison of the structures of the open and closed forms of the protein might provide insights into the molecular events involved in iron binding, additional datasets were collected and the structures were determined by molecular replacment using Molrep.
The structure of wild-type FbpA in the open conformation revealed that the iron atom was co-ordinated by the adjacent tyrosine residues in the C-domain and that there was no phosphate anion present in the anion-binding pocket (Figure 3A). The structure of FbpA in the closed conformation from the same preparation (Figure 3B) had a complete iron co-ordination complex, including the synergistic phosphate anion. These results indicate that the synergistic phosphate anion is not required for binding of ferric ion by the tyrosine residues in the C-terminal domain, and may imply that initial anchoring of the ferric ion is mediated by the tyrosine residues. The phosphate anion clearly stabilizes the co-ordination of the ferric ion and, through interactions with residues in the N-terminal domain, maintainance of the closed conformation. One possible explanation for the presence of phosphate when FbpA crystallized in the closed conformation and its absence when FbpA was present in the open conformation is that there may have been limited phosphate available in the crystallization mixtures and phosphate would strongly favour binding to protein in the closed conformation. Thus any phosphate associated with FbpA in the open conformation that was added to the crystal lattice would tend to equilibrate with free FbpA capable of assuming the closed conformation.
Iron-co-ordination complex of FbpAs, showing Tyr195-OH–Fe and Tyr196-OH–Fe bond distances
Although it was not possible to obtain crystals of mutant proteins in the closed conformation, we decided to determine the structure of the Y195A and Y196A mutant FbpAs to gain insights into the anchoring of the ferric ion by the remaining tyrosine residue. As described in the Supplementary Experimental section, crystals with good diffraction properties were obtained using preparations and conditions similar to those used previously for other FbpA mutants [21,22]. The structures of the iron co-ordination site of the Y195A and Y196A FbpA are illustrated in Figures 3(C) and 3(D) respectively. In both of these structures, the ferric ion is co-ordinated by the remaining tyrosine residue and the synergistic phosphate anion. The presence of this anion may have been biased by the inclusion of exogenous phosphate in the protein preparation used for these crystals . It is salient to note that the OH-Fe bond between Tyr196 and the ferric ion in the Y195A mutant protein is considerably shorter than the OH-Fe bond between Tyr195 and the ferric ion in the Y196A mutant protein. This may explain the higher affinity for iron binding by the Y195A protein (Table 1 and Figure 1).
Eukaryotic transferrins and the ‘bacterial transferrins’, FbpAs from H. influenzae and the pathogenic Neisseria species, have similar means of achieving high-affinity binding of ferric ions, which is essential for their respective physiological roles. The iron-binding site is composed of a similar set of ligands positioned at the base of a cleft between two rigid domains that fluctuate between open and closed conformations, with the iron-loaded form predominantly in the closed conformation and the apo form predominantly in the open conformation [28,29]. The ligands involved in co-ordinating iron are from both the N-terminal and C-terminal domains and thus contribute to maintaining the iron-loaded form in the closed conformation. Structures have been obtained of iron-free lobes of lactoferrin in various states of closure , and mutants of transferrin defective in the iron co-ordinating ligands crystallized in the closed conformation [7,8], suggesting that other interactions contribute to stabilizing the closed conformation. In contrast, structures of mutants in the iron-co-ordinating ligands of FbpA have uniformly been in the open conformation [18,20–22,30] and, despite our best efforts, we were unable to capture mutant proteins in the closed conformation even when seeding with microcrystals in the closed conformation. The shallower cleft and less-extensive domain interface of FbpAs may make the domain closure more dependent upon the ligands co-ordinating the iron and synergistic anion than for transferrins and lactoferrins and may need to be considered when making mechanistic comparisons.
H. influenzae and the pathogenic Neisseria species are capable of using human transferrin as a sole source of iron for growth by an iron-acquisition process mediated by a human transferrin-specific surface receptor . The receptor is composed of a TonB-dependent integral membrane protein, TbpA, and a lipid-anchored surface lipoprotein, TbpB. The receptor binds transferrin, removes iron and transports it across the outer membrane where it is bound by FbpA for subsequent donation of iron to the inner-membrane transport complex. The proposal that binding to the receptor would induce conformational changes (domain separation) in transferrin that lowers the affinity for binding iron  is supported by studies with isolated receptor proteins  or membranes containing receptor proteins  in which transfer of iron to FbpA only occurs in the presence of the receptor proteins. The recent demonstration that apo FbpA preferentially binds to membranes containing TbpA  and that the transport process requires a threshold of iron-binding affinity  suggests that there may be competition for iron between transferrin and FbpA bound to the TbpA protein. Thus delineation of the mechanism of iron removal and transport by the receptor will rely on detailed comparisons between the iron-binding mechanisms of transferrin and FbpA.
Previous studies have demonstrated that Tyr188 in the human transferrin N-lobe is essential for iron binding , suggesting that it may be the residue responsible for initial anchoring of the ferric ion or that there is insufficient flexibility in the iron-binding pocket on the C-terminal domain to accommodate other ligands. In contrast, mutation of either of the adjacent tyrosine residues with a variety of different replacement amino acids does not abrogate iron binding by FbpA (Table 1 and Figure 1). It is interesting to note that the mutants with an alanine residue replacing tyrosine retained the highest affinity for binding iron, suggesting that providing some ‘flexibility’ in the iron co-ordination sphere may be advantageous. Structural studies demonstrate that the ferric ion is bound by the remaining tyrosine residue and synergistic anion in the C-terminal domain (Figures 3C and 3D), indicating that either tyrosine residue can serve as the initial anchoring ligand. These two adjacent tyrosine residues are present in structures from periplasmic iron-binding proteins from a variety of species, including Serratia marcescens, Yersinia enterocolitica , Mannheimia haemolytica  and Campylobacter jejuni , that include a diversity of different modes of iron co-ordination. This indicates that the adjacent tyrosine residues provide a versatile means of anchoring an iron atom with flexibility in accommodating modifications to the iron-co-ordination scheme. Bioinformatics analyses indicate that the ‘twin tyrosine’ motif is widespread among homologues in a diversity of different species and, at present, other modes of iron co-ordination in periplasmic iron-binding proteins have not been identified.
In the present study, we not only show that the Y195A and Y196A mutant FbpAs are capable of binding iron (Table 1 and Figure 1), but also demonstrate that these proteins are capable of mediating transport of iron from the periplasm to cytoplasm in a reconstituted pathway (Figure 2). Thus their apparent inability to mediate iron acquisition from transferrin in a reconstituted pathway in H. influenzae  is probably due to an inability to compete with receptor-bound transferrin for iron as was observed with the H9A and E57A mutants. In the previous study , strains expressing the Y195A and Y196A proteins grew at the same level of exogenous ferric citrate as the strain lacking an FbpA protein, inferring that they were not capable of mediating iron transport. However, in light of the results from the present study, it seems more likely that there are alternate iron-uptake pathways in wild-type H. influenzae that are equally capable of acquiring iron as the FbpABC pathway when the Y195A and Y196A FbpA mutant proteins are present. It is salient to remember that the E. coli reporter strain is defective in iron uptake, which enabled us to detect iron uptake mediated by the FbpABC pathway. This highlights clear advantages that the reporter strain has in monitoring iron acquisition and that there may be value in developing a variety of different reporter strains in different bacterial species.
There is considerable evidence that the synergistic carbonate anion plays a significant role in the initial capture of ferric ion by the C-terminal domain of transferrin and lactoferrin lobes [4,5]. Similarly, the presence of a phosphate anion in the published structures of the apo and iron-loaded forms of the H. influenzae FbpA [12,29] combined with experimental evidence for a strong preference for the phosphate anion  argue for a critical role of the phosphate anion in iron binding. However, the absence of phosphate in the iron-loaded wild-type FbpA in the open conformation (Figure 3) indicates that the presence of phosphate in structural studies may be a function of the conditions of protein preparation and crystallization. In structures containing a synergistic phosphate anion, exogenous phosphate was either added to the samples prior to crystal setups or in buffers used in the purification protocol. In contrast with the kinetic studies that argue for the important role of the synergistic anion, site-directed mutants defective in phosphate binding were fully capable of growth on transferrin iron . Furthermore, although the preparation and crystallization conditions were similar, phosphate was absent from structures of the N175L and N193L mutant proteins while present in the structure of the Q58L mutant, as this residue is on the N-terminal domain. These results, combined with the phosphate-negative open structure obtained in the present study, collectively indicate that, although phosphate may commonly participate in the iron-transport process, it is neither integral nor essential for iron binding or iron acquisition. Considering that these bacteria will be in relatively phosphate-rich environments during invasion or when occupying the submucosal environment of the respiratory tract, but may encounter more-variable conditions when colonizing the mucosal surface, having the flexibility to function without phosphate but operate more efficiently in its presence would be a useful feature for FbpA.
A recent study probing the binding of ferric anion complexes by wild-type and mutant FbpAs from Neisseria gonorrhoeae with time-resolved stopped-flow absorbance and fluorescence spectroscopy demonstrated that the phosphate anion altered the stages and kinetics of binding and led the authors to conclude that initial binding involved ligands in the N-terminal domain . In contrast, our results with the wild-type protein crystallizing in the open conformation (Figure 3A) as well as previous structural studies with mutant FbpAs [20,21,22] strongly argue for the initial binding to the tyrosine residues in the C-terminal domain, as the structures clearly show iron bound to the tyrosine residues with the N-terminal ligands pointing away from the binding site. Our interpretation is that this represents an intermediate stage in iron binding, which requires domain closure to complete the optimal iron-co-ordination complex. We believe that there are alternate interpretations of the kinetic data from the previous study that are consistent with binding to residues in the C-lobe.
In the present study, we obtained structures of iron-loaded wild-type FbpA in the open (Figure 4B) and the closed (Figure 4D) conformations from the same preparation, which could potentially represent different stages in the iron-binding and-release process. The published structure of the apo form of FbpA  (Figure 4B) and a structure lacking the bound phosphate (Figure 4A) represent additional stages of proposed pathways for iron binding under limiting phosphate (1a and 1b, Figure 4) or sufficient phosphate (2a and 2b, Figure 4) conditions.
Models for iron binding by H. influenzae FbpA under low- and high-phosphate conditions
Under phosphate-limited conditions that potentially could exist on the mucosal surface, the apo form of FbpA in the periplasm would not contain the phosphate anion (Figure 4A) and ferric ion released from transferrin by the outer-membrane transport complex would bind directly to the tyrosine ligands in the C-terminal domain (1a, Figure 4B). The transfer of ferric ion from TbpA to FbpA would probably involve a ‘solubilizing’ anion, which might be determined by relative concentrations in the periplasm. Available phosphate would be preferentially bound by the resulting ferric ion–FbpA complex, facilitating completion of the octahedral binding pocket by closure of the two lobes through successive co-ordination of iron by the glutamic acid and histidine ligands and engagement of the glutamic acid in phosphate co-ordination (1b, Figure 4D). The crystallization conditions lacking added phosphate that we used for our structural studies resulted in bound phosphate being present in FbpA in the closed conformation (Figure 4D), supporting the proposal for efficient capture of available phosphate by the Fe–FbpA complex.
Bacteria in the submucosal space or in other body compartments during invasive disease would be exposed to relatively high levels of phosphate such that periplasmic FbpA would contain a bound phosphate anion (2a, Figure 4C). The bound phosphate anion would facilitate binding of ferric ion (2b, Figure 4) and rapid formation of the octahedral iron-co-ordination complex through domain closure (Figure 4D). The process of iron removal by the inner-membrane transport complex is unknown, but the important role that the phosphate anion plays in iron co-ordination might be exploited by displacement of the bound phosphate anion to facilitate the removal of the bound iron atom (Figure 4, reversal of 1b and 1a).
As illustrated in the present study, the relative ease of production of site-directed mutants of FbpAs for biochemical and structural studies, and the ability to test their function in a reconstituted iron-uptake pathway, provides a powerful approach for probing the mechanisms of iron binding and release. However, the inability to crystallize iron-bound mutant proteins in the closed conformation limits the ability to delineate the detailed impact of specific mutations on iron co-ordination. Our results also indicate that one has to be careful about predicting the impact of specific mutations on iron-binding properties and suggest that there may be flexibility in achieving iron binding. We propose that the consistent association of iron with tyrosine residues in the C-terminal domain in structures of wild-type and mutant FbpAs, and the conserved di-tyrosine motif in other periplasmic iron-binding proteins, strongly argues for the critical role of the tyrosine residues in initial anchoring of the ferric ion. More-direct experimental evidence than kinetic analyses  will be needed to support the proposal that iron is initially bound by residues in the N-terminal domain.
Anthony Schryvers was responsible for the development of the research programme, and provided supervision and guidance in implementation of the research and assistance in initial crystallization screens. Trevor Moraes provided direction and assistance in the crystallization experiments and collection of the diffraction data, and was largely responsible for solving the structures of apo and holo wild-type FbpAs. Stephen Shouldice independently solved the structures of the mutant FbpAs. Rong-hua Yu generated most of the site-directed mutations and prepared the wild-type and mutant proteins for crystallization screens. Husain Khambati, with some assistance from Jagroop Singh, subcloned the mutant genes and produced the proteins for spectral analyses, and also prepared mutant genes and performed the pathway-reconstitution experiments; with assistance from Trevor Moraes and Anthony Schryvers, he performed the structural studies with the wild-type FbpA and the attempts at crystallizing mutant proteins in the closed conformation.
We thank Dr Natalie Strynadka (Department of Biochemistry and Molecular Biology, University of British Columbia, Vancouver, BC, Canada) for the use of the X-ray diffractometer, and Hyo Jong Park for her assistance in the purification of the FbpA mutants.
This work was supported by the Canadian Institutes of Health Research [grant number 49603].
The four refined crystal structures of FbpA determined in the present study will appear in the PDB under accession codes 3KN7 (Y195A), 3KN8 (Y196A), 3OD7 (wild-type closed) and 3ODB (wild-type open).