Bordetella pertussis FbpA binds both unchelated iron and iron siderophore complexes.

Bordetella pertussis is the causative agent of whooping cough. This pathogenic bacterium can obtain the essential nutrient iron using its native alcaligin siderophore and by utilizing xeno-siderophores such as desferrioxamine B, ferrichrome, and enterobactin. Previous genome-wide expression profiling identified an iron repressible B. pertussis gene encoding a periplasmic protein (FbpABp). A previously reported crystal structure shows significant similarity between FbpABp and previously characterized bacterial iron binding proteins, and established its iron-binding ability. Bordetella growth studies determined that FbpABp was required for utilization of not only unchelated iron, but also utilization of iron bound to both native and xeno-siderophores. In this in vitro solution study, we quantified the binding of unchelated ferric iron to FbpABp in the presence of various anions and importantly, we demonstrated that FbpABp binds all the ferric siderophores tested (native and xeno) with μM affinity. In silico modeling augmented solution data. FbpABp was incapable of iron removal from ferric xeno-siderophores in vitro. However, when FbpABp was reacted with native ferric-alcaligin, it elicited a pronounced change in the iron coordination environment, which may signify an early step in FbpABp-mediated iron removal from the native siderophore. To our knowledge, this is the first time the periplasmic component of an iron uptake system has been shown to bind iron directly as Fe(3+) and indirectly as a ferric siderophore complex.

Iron is the second most abundant metal in the Earth’s crust and is critical for many life processes despite its extreme insolubility and toxicity under physiological conditions.[1] Thus, biological systems have developed specialized mechanisms for transport, storage, and uptake of this essential nutrient. To obtain iron, many bacteria produce siderophores that bind extracellular iron, diffuse to the bacterial surface, and are taken up via specific receptor proteins.[2−6] Certain bacteria can also utilize siderophores produced by other microbial species (xeno-siderophores) if they produce the cognate transporters. Heme can also serve as an iron source,[7,8] and some bacteria can utilize host transferrin- or lactoferrin-bound iron.[9−12] In Gram-negative bacteria, iron sources are typically bound by an outer membrane receptor (OMR) and transported to the periplasm using energy supplied by the TonB complex.[13] A substrate-specific periplasmic binding protein (PBP) interacts with the iron source, binds the cargo between two lobes, and delivers it across the periplasm to the appropriate ATP-binding cassette (ABC) transporter for transit to the cytoplasm.[14−16]

The PBPs involved in bacterial iron uptake have historically been classified as either unchelated iron binders or chelated iron binders (i.e., iron-siderophore complex binders).[14,17−19] However, the unchelated iron binders are not always categorized together. (Unchelated iron has variously been referred to in the literature as naked or inorganic iron in reference to direct binding to the +3 ion.) One prototypical unchelated iron binder is the periplasmic ferric iron binding protein FbpANg from Neisseria gonorrheae.[20] FbpANg is required for utilization of host-transferrin-derived iron and binds Fe3+ directly in the first coordination shell in the presence of a requisite synergistic anion.[10,19−24] Unchelated iron binding proteins from other bacteria, including some that do not produce transferrin or lactoferrin receptors, have also been shown to bind iron directly. However, the anion-dependence, iron-coordination geometry, and type of amino acid side chains involved in iron sequestration vary.[14,25−28] It has been shown that even among homologues of FbpANg, direct iron binding is accomplished in a variety of ways by bacteria that use a variety of iron sources.[26] On the other hand, one prototypical binder of chelated iron is the iron-siderophore binding PBP FhuD from Escherichia coli.[29−31] The reported crystal structures of FhuD show that the first coordination shell of iron is occupied by the siderophore, and the protein occupies the second coordination shell.[29]

The schemes currently used to categorize protein families, of which the periplasmic components of iron uptake systems are members, use a variety of classification criteria, but all imply separate functions: either unchelated iron transport or iron-siderophore transport.[7,18]Here we report a novel capacity for a periplasmic protein that is involved in iron uptake in Bordetella pertussis: the ability to bind iron directly and the ability to bind iron-siderophore complexes.

The Gram-negative bacterial species B. pertussis is a respiratory pathogen that causes whooping cough in humans.[32,33]B. pertussis can acquire iron using its heme uptake system[34] or its native dihydroxamate siderophore alcaligin (ALC; Figure S1 in Supporting Information [SI]).[35,36] It can also use xeno-siderophores, including desferrioxamine B (DFB), ferrichrome (FC),[37] and enterobactin (ENT), among others (Figure 1; Figure S1 in SI).[38] Of the Bordetella siderophore systems, only the ALC and ENT systems have been characterized.[39,40] The ALC genes encode ALC biosynthesis, export and regulation activities, and the OMR.[39] The ENT gene cluster encodes the OMR, an IroE hydrolase orthologue, and a transcriptional regulator.[40] Notably, there are no genes predicted to encode cognate periplasmic iron-siderophore binders (or ABC transporters) within the ALC and ENT chromosomal gene clusters.

Figure 1

Bordetella FbpA-dependent iron uptake via alcaligin (pink rectangles) and xeno- (e.g., ENT, DFB, FC; green circle) siderophores. Host iron sources include transferrin or lactoferrin (Tf/Lf) as well as others not shown. TonB-dependent receptors (green and pink) are shown in the outer membrane (OM); the TonB complex and the ABC transporter FbpB/CBp are in the cytoplasmic membrane (CM). FbpABp is shown in the periplasm, and a previous study showed that it is important in unchelated iron and iron-siderophore utilization.[42]

Recently, we identified fbpA as an iron repressible B. pertussis gene that is also present in other Bordetellae.[42]fbpA encodes a periplasmic protein (FbpABp) with significant structural similarity to iron binding proteins from Serratia marcescens,Neisseria spp., and Haemophilus influenzae.[13,16,17]B. pertussis fbpA is predicted to be cotranscribed with fbpB and fbpC, which encode the permease and ATPase components of an ABC transporter. B. pertussis fbpA mutants were defective in using unchelated iron for growth and, remarkably, were also defective in using the native siderophore and xeno-siderophores for growth.[42] Heme iron assimilation was not affected by fbpA inactivation. Since the prototypic bacterial siderophore system relies on a siderophore-specific PBP and associated ABC transporter, these Bordetella results were unusual as ALC, ENT, DFB, and FC are quite structurally different (Figure S1 in SI). Not only are their functional groups different, but the hexadentate siderophores ENT, DFB, and FC bind iron in a 1:1 ratio, whereas tetradentate ALC binds iron in a 2:3 ratio (Fe3+:ALC) at physiological pH (see SI).[4,36,43,44] These results suggested that FbpABp interacts with diverse ferric siderophores to mediate iron transport to the cytosol. To our knowledge, a PBP capable of binding iron directly and capable of binding iron-siderophore complexes has not been described previously. The biological purpose of direct interaction with iron by a PBP involved in iron-siderophore utilization may represent a new paradigm for PBP function.

The X-ray crystal structures of apo-FbpABp (PDB: 1Y9U), Fe–FbpABp–carbonate (PDB: 2OWT), and Fe–FbpABp–(oxalate)2 (PDB: 2OWS) have been reported and show that FbpABp can directly coordinate Fe3+ through its conserved tyrosine residues (Y143, Y199, and Y200).[26,27] Similar to FbpANg,[20] the crystal structures of FbpABp show iron bound directly by the protein with an anion present.[26,27] These solid state studies of FbpABp, coupled with its known requirement for ferric siderophore utilization,[26,27,42] indicate that FbpABp is similar to other unchelated iron binders, but also has potentially unique functions in ferric siderophore utilization.

In this study, using FbpABp produced naturally in Bordetella cells, we quantified Fe3+ sequestration by FbpABp in the presence of various synergistic anions in solution. We also demonstrated that FbpABp binds ferric complexes of ALC, ENT, DFB, and FC, all with micromolar affinity, and characterized those binding interactions in solution. This report describes B. pertussis FbpABp as the first known example of a promiscuous periplasmic iron transport component that binds both unchelated iron and ferric siderophores and thus may act as a nodal point in non-heme Bordetella iron transport.

Experimental Methods

Materials

All chemicals used were of highest purity grade: MES (2-(N-morpholino)ethanesulfonic acid) buffer, desferrioxamine B mesylate salt, and ferrichrome solution (iron free, from Ustilago sphaerogena) (Sigma Chemicals); sodium chloride, sodium carbonate (Fisher Chemical); sodium oxalate (Allied Chemical); sodium citrate (J. T. Baker); and sodium NTA (Aldrich Chemical Co.). Iron-free 50 mM MES, 100 mM NaCl buffer solutions were prepared in acid-washed buffer bottles by dissolving the required amount of MES and NaCl in deionized water, and the pH was adjusted to 6.5 by addition of acid/alkali. Chelex 100 (BioRad) beads were added to this solution to remove iron contamination and left overnight followed by filtration to remove the beads. All experiments were performed in pH 6.5 buffer as the pH of the periplasm is believed to be slightly acidic relative to the surroundings.[45,46] Apo-alcaligin and enterobactin were purified as described previously, from supernatant fluids of Bordetella bronchiseptica and E. coli, respectively.[36,41,47]

Purification of FbpABp

B. bronchiseptica strain BRM72 (ΔfbpA) carrying plasmid pBBR/fbpA encoding fbpA from B. pertussis strain Tohama I was used to produce the FbpABp protein in the periplasm under natural iron starvation conditions. Bacteria were cultured in iron-depleted Stainer–Scholte medium,[48] containing 0.5% casamino acids and gentamicin (10 μg/mL) at 35 °C. FbpABpwas extracted from the bacteria by osmotic shock treatment as described previously.[49] The osmotic shock fraction was clarified by centrifugation, followed by filtration using a low protein binding polyethersulfone membrane filter with 0.2 μm pore size. The clarified osmotic shock fraction was dialyzed against 10 mM Tris buffer, pH 8.6, at 4 °C, and applied to a 2.5 cm × 15 cm DEAE CL-6B Sepharose anion-exchange column. The column was washed with 10 bed volumes of 10 mM Tris buffer, pH 8.6, then eluted with a 4-bed volume continuous gradient of 0.0–0.5 M NaCl. Peak fractions were pooled and dialyzed against 10 mM Tris buffer, pH 8.6. Purified FbpABp was concentrated by ultrafiltration and the protein concentration was determined using the Bradford method with bovine serum albumin as the standard.[50] The identity of the FbpABp protein was confirmed by liquid chromatography–tandem mass spectrometry (80% coverage, 100% protein identification probability) using an LTQ ion trap mass spectrometer. An SDS-PAGE gel showing FbpABp is presented in the SI as Figure S10.

Preparation and Characterization of Fe-FbpABp–X Assemblies

UV–vis spectra were recorded using a Cary-50 UV–vis spectrophotometer with 1 cm path length cells in 50 mM MES, 100 mM NaCl at pH 6.5 and room temperature. The molar absorptivity of holo-FbpABp was calculated on the basis of the absorbance band at 280 nm from an iron-loaded FbpABp sample (ε = 44000 M–1 cm–1) and the characteristic absorbance band in the visible region of the spectrum in the presence of various anions. The molar absorptivity of apo-FbpABp is 38500 M–1 cm–1 at 280 nm and is ∼10% lower than that of holo-FbpABp following a trend reported by us for FbpANg in a previous study.[23]

Fe–FbpABp–X (X = carbonate or oxalate) was prepared by adding 50 equiv of Na-carbonate or oxalate from freshly prepared stock solutions (50 mM) to apo-FbpABp in 50 mM MES, 100 mM NaCl at pH 6.5. The resulting solutions were allowed to equilibrate for 30 min before 10 equiv of standardized Fe(ClO4)3 solution were added, and the solutions were gently mixed for 30 min and then allowed to stand overnight at 4 °C to precipitate any unreacted iron, which was removed by filtration with a syringe-driven 0.2 μM filter unit (Corning Corporation). Fe–FbpABp–citrate was prepared by adding 10 equiv of Na-citrate from a freshly prepared stock solution (20 mM) to apo-FbpABp followed by 2 equiv of Fe-citrate (1:1, 2 mM). Similarly, to prepare Fe–FbpABp–NTA, 10 equiv of Na-NTA from a stock solution (20 mM) were added to apo-FbpABp followed by the addition of 2 equiv of Fe-NTA (1:1, 2 mM). Additional Fe(ClO4)3/Fe-citrate/Fe-NTA was titrated into the solutions of protein complexes until no further increase in intensity of the corresponding charge transfer band was observed, ensuring fully iron-loaded protein complex formation. Formation of the corresponding ternary complexes was confirmed by the appearance of well-resolved charge transfer bands in the visible region of the absorption spectrum of the respective Fe–FbpABp–X species (X = citrate, carbonate, oxalate, or NTA). All protein complexes were dialyzed against 50 mM MES, 100 mM NaCl at pH 6.5 buffer three times using Slide-A-Lyzer dialysis cassette (Thermo Scientific) to remove excess iron and anion.

A competitive equilibrium between Fe–FbpABp–X (X = oxalate, carbonate, citrate, or NTA) and EDTA (eq 1) was employed to determine the Fe3+ binding constants of ternary complexes for Fe–FbpABp–X, as defined in eq 2, using Experimental Methods described previously.[23]

Characterization of apo-FbpABp Interactions with Ferric Siderophores

Fluorescence spectroscopy was used to quantitatively determine the interaction between apo-FbpABp and ferric siderophores. All fluorescence spectra were recorded on JOBIN-YVON-SPEX Fluorolog3 fluorimeter at right angle mode with 5 nm slit width and in 50 mM MES, 100 mM NaCl buffer at pH 6.5 and room temperature. To minimize the intrinsic emission from the tyrosine residues, the apo-FbpABp was excited at 297 nm. The observed intrinsic emission from apo-FbpABp is the summation of emissions from individual tryptophan residues and is centered at 352 nm under our experimental conditions. This indicates a hydrophilic environment on average for the four tryptophan residues present in FbpABp.[51,52] Quenching of the intrinsic tryptophan emission band upon ligand (Fe3+-siderophore) addition is an indication of reduction in the concentration of the free apo-FbpABp in solution that can emit at 352 nm, and these data were used to determine the binding affinity of FbpABp with various Fe3+-siderophores. An apparent Fe3+-siderophore/apo-FbpABp dissociation constant (Kd) corresponding to the equilibrium in eq 3 was determined by fluorescence titration using eq 4, where P represents FbpABp, Fe-sid represents both ferric native and ferric xeno-siderophores and P(Fe-sid) represents the FbpABp(Fe-siderophore) complex. Reproducible results were only obtained for a one-site binding model. The experimental design and data analysis approach was the same asthat used previously by us and others.[22,31,52,53]Kd values were obtained from plots of eq 5, where Q% is the observed % quenching at each point in the titration and Qmax is the maximum % quenching.[22]

A possible exchange of Fe3+ between Fe-siderophores and apo-FbpABp was monitored using UV–visible spectroscopy by monitoring the CT band of the Fe-siderophore over a 6 h time period (protein:ferric xeno-siderophore 1:2; protein:ferric native siderophore = 2:1).

Protein Modeling

Iron and siderophore binding sites were modeled starting with RCSB Protein Data Bank structures 1Y9U:A and 2OWT (to which hydrogens were added).[26]1Y9U is the structure for the B. pertussis ferric binding protein in the apo (open) conformation. 2OWT represents the holo (closed) structure, and its ligands, CO32– and iron, were removed before docking. Ligands for docking were downloaded from either RCSB Protein Data Bank (www.rcsb.org) or, in the case of ferrioxamine B[54] and ferric-alcaligin (Fe2ALC3),[43] and the alcaligin monomer (ALC), from the Cambridge Crystallographic Data Centre (reference codes: 155586, TEQKUV, and C00080, respectively). Patchdock[55] was used to find the initial docking sites, and models were fine-tuned with Chimera.[56] The superposition of the alcaligin (monomer)-bound model with the 2OWS structure was created with the Matchmaker utility implemented in Chimera. Models were evaluated using ERRATv2,[57] and the best scoring value for each is reported.

Results and Discussion

Crystal structures of holo-FbpABp established that it binds Fe3+ directly in the presence of synergistic anions (oxalate and carbonate),[26,27] and Bordetella growth studies showed that FbpABp is essential for the utilization of Fe3+ bound to multiple structurally distinct siderophores.[42] The flexibility of the β-sheet hinge region of FbpABp is consistent with the possibility of such substrate binding promiscuity. Modeling of ferric native and xeno-siderophore binding between the two lobes of FbpABp (Figures S2–S5 in SI) predicted favorable binding interactions, supporting the possibility that FbpABp can act as a ferric siderophore chaperone. In the following sections we describe the interaction of FbpABp with unchelated Fe3+ and with iron-siderophore complexes utilized by Bordetella.

FbpABp Strongly Sequesters Unchelated Fe3+ in the Presence of Various Synergistic Anions

Crystal structures reported for Fe–FbpABp–(oxalate)2 and Fe–FbpABp–carbonate show that Fe3+ is bound to the three conserved tyrosine residues on the C-lobe of the protein.[26,27] Until now there has been no report that quantifies FbpABp–Fe3+ binding interactions. Here we have determined the Fe3+ affinity of each Fe–FbpABp–X assembly, where X is citrate, carbonate, oxalate, or nitrilotriacetate (NTA). Figure 2 shows a representative overlay of the visible spectra for all Fe–FbpABp–X assemblies prepared individually as described in Experimental Methods, indicating the anion promiscuity of FbpABp in sequestering Fe3+ and modulation of the position of the charge transfer (CT) bands by various anions (Table 1). No Fe3+ sequestration was observed in the absence of a suitable synergistic anion. In silico modeling also identified an arginine residue (R105) involved in anion binding in other FbpABp homologues (Table S1 in SI) that is conserved in FbpABp. The appearance of characteristic CT bands that are unique for different anions indicates that these anions directly participate in iron sequestration at the first coordination shell and are synergistic anions.

Figure 2

Visible region spectra of Fe-FbpABp-X (X = citrate, carbonate, oxalate, or NTA) in 50 mM MES, 100 mM NaCl at pH 6.5. Protein concentrations are indicated.

Table 1

Biophysical Characterization of Fe-FbpABp-X (X = Citrate, Carbonate, Oxalate, or NTA) at pH 6.5 in 50 mM MES, 100 mM NaCl

anion	λmax, nm	ε, M–1 cm–1	K′eff, M–1a
citrate	410	3433	7.0 ± 3.7 × 1016
NTA	466	3049	4.7 ± 2.0 × 1016
oxalate	440	3143	5.5 ± 1.2 × 1016
carbonate	412	3390	1.2 ± 0.03 × 1016

Effective stability constants corresponding to eq 2. Values reported represent the variation in two independent determinations.

The observed Fe3+ affinity constants (K′eff) defined in eq 2 and determined as described in Experimental Methods are listed in Table 1. K′eff values (∼1016 M–1) are approximately 1–2 orders of magnitude less than those reported for FbpANg, which may be due to a difference in coordination environment provided by these two proteins.[23,26,27,58]

We conclude that FbpABp is capable of sequestering Fe3+aq with high affinity (K′eff ≈ 1016 M–1) in the presence of various synergistic anions. In these assemblies, the anions and the protein both occupy the first coordination shell of Fe3+.

FbpABp Binds Ferric Siderophore Complexes

Our hypothesis based on Bordetella growth experiments using fbpA mutants42 and in silico modeling (Figures S2–S5 in SI) proposed that FbpABp has the potential to bind ferric native and xeno-siderophores as illustrated in eq 3. To investigate this possible interaction we carried out fluorescence quenching titrations involving apo-FbpABp and ferric-alcaligin, or one of the ferric xeno-siderophores. Apo-FbpABp, with four tryptophan residues in its structure, emits at 352 nm when excited at 297 nm, indicating a hydrophilic average environment for the tryptophans.[51] Although both ferric native and xeno-siderophores showed concentration-dependent quenching of the intrinsic emission band, the concentration dependence, change in band position, and maximum quenching (Qmax) differed appreciably between the two types of siderophores. This suggests a differential mode of interaction between FbpABp and the ferric native siderophore versus the ferric xeno-siderophores. The affinities (Kd; eq 3) of all the holo-siderophores for FbpABp were calculated using these concentration dependent quenching data as described in Experimental Methods. Figure 3a shows representative quenching spectra when Fe2(ALC)3 was titrated into a solution of apo-FbpABp, and Figure 3b shows the fit of these data using a one site binding model.

Figure 3

(a) Representative fluorescence emission spectra for FbpABp (1 μM) in the presence of increasing concentrations of Fe2(ALC)3 (0–2 μM). Upon addition of Fe2(ALC)3, emission decreases and a blue shift of ∼8 nm is observed. Refer to the Experimental Methods section for experimental conditions. (b) A plot of Q% FbpABp fluorescence emission (352 nm) vs increasing [Fe2(ALC)3]. Filled circles represent actual data points, and the smooth line is the best fit of eq 5 to the data. Average Kd for eq 3 = 0.13 ± 0.04 μM and Qmax = ∼80%.

The Kd value for Fe2(ALC)3 binding to apo-FbpABp is 0.13 ± 0.04 μM. Similar micromolar affinity for ferric-hydroxamate siderophore binding to FhuD from E. coli has been reported.[31] Figure 3a shows an ∼8 nm blue shift in the emission band, indicating a change in average tryptophan environment from hydrophilic to hydrophobic surroundings.[51] Consequently, we conclude that addition of Fe2(ALC)3 to apo-FbpABp leads to a change of conformation of FbpABp concomitant with a change in tryptophan hydrophilicity.

Kd values calculated using the same mathematical model for the ferric xeno-siderophores (ENT, DFB, and FC) were found to be an order of magnitude higher than the Kd calculated for Fe2(ALC)3 (Kd ≈ 1 μM). Representative data are presented in SI (Figures S6–S8). The Qmax for the xeno-siderophore titrations range from 20 to 60%, whereas for the Fe2(ALC)3 titration Qmax is 80%. We interpret this as indicative of the involvement of the tryptophan residues (or a subset) to a greater extent for the protein–ligand interaction in the presence of Fe2(ALC)3 than for the ferric xeno-siderophores.

When the apo-siderophores (ALC and ENT, as representative hydroxamate and cathecholate siderophores, respectively) were titrated into a solution containing apo-FbpABp in separate experiments (final protein:apo-siderophore = 1:4 and 1:2, respectively), the apo-protein fluorescence emission band did not show significant quenching (<16%; Figure S9 in SI). This suggests that only a very weak interaction occurs between these apo-siderophores and the apo-protein. A similar result was obtained when a UV–vis spectral difference experiment was performed between apo-DFB and apo-FbpABp (results not shown), once again indicating that the apo-protein is not able to bind an apo-siderophore effectively. These experiments suggest that the “binding interaction” that we observe quantitatively using a fluorescence emission quenching assay for holo-siderophores and apo-FbpABp is ligand-conformation specific and in fact requires the siderophore to be bound to Fe3+.

These results indicate that in addition to binding directly to unchelated Fe3+, FbpABp can bind the native ferric siderophore (ALC) with Kd = ∼0.1 μM and various ferric xeno-siderophores with Kd = ∼1.0 μM. This finding is consistent with earlier growth studies demonstrating that Bordetella fbpA mutants are impaired in their utilization of ferric siderophores as well as unchelated iron.[42] These results also support our hypothesis derived from in silico modeling (Figures S2–S5 in SI) and represent the first quantification of the capacity of a bacterial periplasmic protein to bind iron both directly and indirectly.

Fe3+ Exchange within the FbpABp/holo-Siderophore Assembly

The fact that FbpABp can bind unchelated Fe3+ with high affinity (Table 1) and can form a “chaperone-like” assembly with a ferric siderophore complex with micromolar binding affinity (eq 3) led us to expand our working hypothesis to include the possibility of an exchange mechanism for Fe3+ between the protein bound holo-siderophore and the apo-protein. This is illustrated in eq 6, where “sid” indicates both native and xeno-siderophores, and Y represents either a siderophore or an opportunistic environmental synergistic anion. Such a process may represent a key step in the iron transport mechanism to the cytoplasm.

To test this extended hypothesis we performed time-dependent UV–vis spectroscopy experiments in which apo-FbpABp was added to ferric siderophore complexes, and the characteristic CT band for the ferric siderophore was monitored over time. Exchange of Fe3+ from the siderophore into the FbpABp binding site will change the position/intensity of the ferric siderophore CT band due to a donor group change in the first coordination shell of the Fe3+.

For all ferric xeno-siderophore complexes investigated, the characteristic ligand-to-metal charge transfer band remained constant in position and intensity, even after exposure to apo-FbpABp for 3 h at pH 6.5 (Figure 4a). These results indicated that there is no change of the ferric siderophore first coordination shell in the presence of the protein and discounts an exchange reaction as described in eq 6. This observation is not surprising, given that the stability constants for the holo-siderophore complexes (∼1023–1049 M–1),[3,4] are much higher than the Fe3+ association constant observed for holo-FbpABp (∼1016 M–1) (Table 1). However, a conformational change destabilizing the iron-siderophore binding pocket in the presence of a small-molecule chelator could facilitate dissociation of the Fe3+ into the protein binding pocket. In order to test this hypothesis for the iron-exchange reaction shown in eq 6, we carried out the reaction in the presence of biologically relevant anions. Our choice of anions was citrate and phosphate, which either form stable Fe–FbpABp–X complexes or have been shown to facilitate the insertion of Fe3+ into N. gonorrheae FbpA, respectively.[59] Even in the presence of 50-fold excess of either of these anions there was no exchange of Fe3+ between a xeno-siderophore and FbpABp. On the basis of these observations we conclude that FbpABp is unable to remove Fe3+ from a ferric xeno-siderophore or detectably alter the first coordination shell of Fe3+ under the in vitro conditions used.

Figure 4

(a) UV–vis spectra with charge transfer (CT) band at 493 nm for Fe-ENT (15.4 μM) before and after addition of FbpABp (7.5 μM) in 50 mM MES, 100 mM NaCl at pH 6.5; (b) UV–vis spectra for Fe2(ALC)3 (9 μM) before and after addition of apoFbpABp (18 μM) showing a shift in λmax from 428 nm (CT band for Fe2ALC3) to 410 nm. The mixture of Fe2(ALC)3 and FbpABp was monitored for 3 h, and the spectrum displayed here is from the end of that incubation.

The results were quite different for the reaction between apo-FbpABp and holo-ALC, the native siderophore produced and utilized by the Bordetellae. At our experimental conditions the predominant species for holo-ALC is Fe2(ALC)3 (see SI), which is the expected in vivo form of the ferric siderophore and is hypothesized to be the form of the complex recognized by the outer membrane receptor FauA.[36,44,60] The ligand-to-metal CT band characteristic of Fe2(ALC)3 occurs at 428 nm, whereas the mononuclear species Fe(ALC)(OH2)2+, which predominates at pH < 1, absorbs at longer wavelengths, λmax = 500 nm.[44] Immediately after Fe2(ALC)3 was mixed with a solution of apo-FbpABp the λmax shifted from 428 to 410 nm (Figure 4b). As reported in Table 1, Fe–FbpABp–carbonate exhibits an absorption peak at 412 nm. The possibility of forming a carbonate complex was eliminated by the observation that this 410 nm peak appeared even in the absence of carbonate (degassed sample of Fe2(ALC)3 was mixed with degassed apo-FbpABp under an argon blanket). The appearance of this new absorption peak can be interpreted as due to a perturbation of the first coordination shell of Fe3+ in Fe2(ALC)3 as a result of binding with FbpABp. This change of environment for Fe3+ could simply be due to a binding interaction between the protein and the holo-siderophore or could be indicative of the first step of iron release from Fe2(ALC)3, which requires reorganization of the Fe(III) first coordination shell. As noted above, the fluorescence emission quenching experiment involving apo-FbpABp and Fe2(ALC)3 also shows a lower Kd value, and a movement of the tryptophan residues from hydrophilic to hydrophobic environment (Δλemission ≈ 8 nm; Figure 3a) compared to xeno-siderophore and apo-FbpABp interactions. These changes in the UV–vis and fluorescence spectral band positions and the greater observed affinity of FbpABp for the native siderophore (compared to the affinity with xeno-siderophores) suggest that reaction of Fe2ALC3 with apo-FbpABp may result in the formation of a precursor complex to Fe-FbpABp-ALC under our experimental conditions.

Our modeling study shows the feasibility of the formation of Fe–FbpABp–ALC as shown in Figure 5a. In our model ALC serves as the “synergistic anion” in the exchange of Fe3+ from Fe2ALC3 to FbpABp. This may represent a step in the process of delivering iron to the cytosol. In Figure 5b we have superimposed the model of Fe–FbpABp–ALC on the reported crystal structure of Fe–FbpABp–(oxalate)2 (PDB: 2OWS). Both the model and the crystal structure show Fe3+ sequestered by two conserved tyrosine residues from FbpABp and the rest of the octahedral geometry first coordination shell occupied by an “external ligand” or synergistic anion (either ALC or two oxalate anions). The model fits very well with the crystal structure (RMSD 0.578 Å) and shows that in the model the ALC backbone can interact with the same amino acid residues that interact with the two oxalates in the crystal structure (Arg137 (2.67 Å); Asp179 (4.23 Å); Tyr200 (2.42 Å), Tyr143 (2.79 Å)). Two additional anion–protein interactions are predicted by the model of Fe–FbpABp–ALC (Glu15 (1.42 Å) and His141 (4.25 Å)). This superposition between the modeled structure and the crystal structure of 2OWS strongly supports the feasibility of forming a Fe–FbpABp–ALC assembly and may explain the appearance of the 410 nm band in our Fe3+-exchange experiment.

Figure 5

Panel (a): Model of Fe-ALC in the binding cleft of FbpABp (2WOT); ERRAT score 95.425. Docked structure created with Patchdock using 2OWT as receptor and the ferric-alcaligin monomer (obtained from CCDC: C00080) as ligand. Panel (b): Model of Fe–FbpABp–ALC superimposed on the crystal structure of Fe–FbpABp–(oxalate)2 (2OWS); 2OWS is shown in orange, the two oxalate anions are red, FbpABp is tan, and alcaligin is blue. Structures were aligned using the “Matchmaker” utility in Chimera. RMSD: 0.578 Å.

Conclusion

The results presented here demonstrate a previously undescribed dual capacity for a periplasmic iron-uptake protein. FbpABp is capable of tightly binding unchelated Fe3+ directly in the presence of synergistic anions in vitro (K′eff ≈ 1016 M–1) and can also bind the native ferric siderophore (Fe2(ALC)3) and ferric xeno-siderophore complexes with 0.1 and 1.0 μM affinities, respectively. Furthermore, Fe2(ALC)3 binding by FbpABp perturbs the coordination of Fe3+ in the native alcaligin siderophore complex in such a manner as to indicate partial dislocation of Fe3+ to the FbpABp binding site. This observation suggests a possible iron removal mechanism. These novel abilities are consistent with Bordetella studies showing that FbpABp is required by these respiratory pathogens for the utilization of both inorganic iron and siderophore-bound iron.[42] Our report describes the unique and opportunistic binding modes of Bordetella FbpA, hints at the possible function of this promiscuity, and begs the question, can other PBPs involved in iron-uptake bind iron both directly and indirectly?