The Staphylococcus aureus Siderophore Receptor HtsA Undergoes Localized Conformational Changes to Enclose Staphyloferrin A in an Arginine-rich Binding Pocket

Abstract

Staphylococcus aureus uses several efficient iron acquisition strategies to overcome iron limitation. Recently, the genetic locus encoding biosynthetic enzymes for the iron chelating molecule, staphyloferrin A (SA), was determined. S. aureus synthesizes and secretes SA into its environment to scavenge iron. The membrane-anchored ATP binding cassette-binding protein, HtsA, receives the ferric-chelate for import into the cell. Recently, we determined the apoHtsA crystal structure, the first siderophore receptor from Gram-positive bacteria to be structurally characterized. Herein we present the x-ray crystal structure of the HtsA-ferric-SA complex. HtsA adopts a class III binding protein fold composed of separate N- and C-terminal domains bridged by a single α-helix. Recombinant HtsA can efficiently sequester ferric-SA from S. aureus culture supernatants where it is bound within the pocket formed between distinct N- and C-terminal domains. A basic patch composed mainly of six Arg residues contact the negatively charged siderophore, securing it within the pocket. The x-ray crystal structures from two different ligand-bound crystal forms were determined. The structures represent the first structural characterization of an endogenous α-hydroxycarboxylate-type siderophore-receptor complex. One structure is in an open form similar to apoHtsA, whereas the other is in a more closed conformation. The conformational change is highlighted by isolated movement of three loops within the C-terminal domain, a domain movement unique to known class III binding protein structures.

Introduction

Iron is an essential component of many biological systems playing roles in most forms of life (1). Despite its abundance in biological systems, free iron is scarce in most environments, especially those encountered by pathogenic bacteria (2, 3). However, iron-complexes are abundant in the human body, and bacterial pathogens use multiple distinct strategies to acquire these iron stores. Iron in the human body is primarily located within iron-shuttling glycoproteins, such as transferrin, in the iron storage protein, ferritin, and as the central atom in heme moieties from numerous proteins, including hemoglobin and cytochromes (4). Because the human body contains sufficient amounts of complexed iron to support bacterial growth, the ability to utilize these iron sources offers a significant selective advantage to a pathogen.

Staphylococcus aureus is one of the most commonly acquired bacterial infectious agents in hospitals and a significant source of infection in community acquired infections of medical concern (5). Its prevalence combined with escalating antibiotic resistance within hospital and community isolated strains has made understanding S. aureus pathogenesis pertinent to combating infection (6). This successful bacterial pathogen invades and causes disease in most human body tissues due to its enormous repertoire of virulence factors (7–9).

Iron uptake systems are paramount to survival in any environment, but the diversity of S. aureus iron uptake systems likely contributes to its ability to thrive in most mammalian tissues. S. aureus possesses the iron-regulated surface determinant heme transport system (10). This system has been well characterized, and ligand-bound structures are determined for most components of the system (11–16). S. aureus can also grow on holo-transferrin or lactoferrin as an iron source, which is likely mediated by specific binding on the cell surface (17, 18). S. aureus is able to synthesize and utilize siderophores, small iron chelating molecules (for a recent review of siderophore uptake in S. aureus, see Ref 19). Uptake of exogenous hydroxamate-type siderophores occurs via the Fhu transport machinery (20–25). S. aureus synthesizes two siderophores of known structure: staphyloferrin A (SA)⁴ and staphyloferrin B (SB) (26–30). Biosynthesis and transport of SB in S. aureus is mediated by proteins encoded by the sbn (biosynthesis) and sir (transport) operons (31). Independent work from two groups identified the locus encoding the SA biosynthetic enzymes (32, 33). Beasley et al. (32) further demonstrate that htsABC, situated adjacent to the SA biosynthetic locus in the S. aureus genome, encodes an ABC transporter required for ferric-SA (Fe-SA) import. HtsABC was previously implicated in heme transport (34), and the possibility of a dual role in heme uptake has not been excluded.

Recently, we presented the crystal structure of apoHtsA (32), the first structure of a siderophore receptor protein from a Gram-positive bacterium. The structure is a two-domain protein with a single α-helix bridging the two domains, similar to related metal and metal compound ABC transporter-associated binding proteins (35, 36). Herein, we describe structures of the HtsA-Fe-SA complex from two different crystal forms. The structures are the first for an α-hydroxycarboxylate-type siderophore-receptor complex and provide insights into ligand recognition and the conformational change required for productive interaction with the permease (HtsBC).

EXPERIMENTAL PROCEDURES

Cloning and Protein Expression

Recombinant HtsA was produced by cloning htsA from S. aureus Newman genomic DNA into pET28a. The HtsA expression construct was designed to exclude the N-terminal secretion signal and lipidation site (residues 1–20) and 17 additional N-terminal residues (21–37) that were omitted from the design because of predicted disorder (37). pET28a-htsA (to express residues 38–327) was transferred into Escherichia coli strain ER2566. Cultures were grown in 2× YT media at 30 °C to an optical density of ∼0.8. At this point the culture temperature was shifted to 25 °C, and protein expression was induced with 0.4 mm isopropyl 1-thio-β-d-galactopyranoside. Induced cells were incubated for an additional 16 h. Cells were resuspended in 50 mm Tris, pH 8.0, 100 mm NaCl and disrupted using an EmulsiFlex-C5 homogenizer (Avestin). His₆-tagged protein was purified using a His-Trap HP column (GE Healthcare) in the same resuspension buffer and eluted with a 0–500 mm imidazole gradient. His₆-HtsA was dialyzed into 50 mm HEPES, pH 7.8, and the His₆ tag was cleaved by thrombin digestion (1:500 mass ratio HtsA: thrombin). HtsA was further purified by cation exchange chromatography (Source 15S, GE Healthcare) in 50 mm HEPES, pH 7.8, and eluted with a 0–500 mm NaCl gradient. Protein samples were dialyzed into 20 mm Tris, pH 8.0, for crystallization.

Recombinant SfaB and SfaD (SA synthetases) were required to produce staphyloferrin A in vitro. The sfaB and sfaD coding regions were amplified from S. aureus Newman genomic DNA and cloned into pET28a(+) for overexpression in E. coli BL21 (DE3). Cultures were grown in Luria broth (Difco) supplemented with 30 μg/ml kanamycin at 37 °C to an optical density of ∼0.9. Isopropyl 1-thio-β-d-galactopyranoside (500 μm) was added, and cultures were incubated an additional 18 h at room temperature with shaking. Cells were harvested by centrifugation at 15,000 × g and resuspended in binding buffer consisting of 50 mm HEPES, pH 7.4, 500 mm NaCl, 10 mm imidazole and passed through a French pressure cell at 1500 p.s.i. Cell lysate was centrifuged at 15,000 × g to remove unbroken cells and debris before the supernatant was subjected to additional centrifugation at 150,000 × g for 60 min to precipitate the insoluble material. The soluble fraction was then applied to a 1-ml HisTrap nickel affinity column (GE Healthcare) equilibrated with binding buffer, and the His₆-tagged proteins were eluted with a gradient of 0–80% elution buffer over 20 column volumes (elution buffer consisted of 50 mm HEPES buffer, pH 7.4, 500 mm NaCl, 500 mm imidazole). Proteins were then dialyzed into 50 mm HEPES, pH 7.4, 150 mm NaCl, and 10% glycerol at 4 °C. Protein purity was confirmed using sodium dodecyl sulfate-polyacrylamide gel electrophoresis and then frozen (at −80 °C) and stored as 100-μl aliquots. The protein yields for SfaB and SfaD were 6.8 and 25 mg/liter of culture, respectively.

Staphyloferrin A Enrichment from Culture Supernatant

Staphyloferrin A was enriched from concentrated S. aureus culture supernatants as previously described (32). For use in crystallization, FeCl₃ was added to concentrated siderophore extracts to a final concentration of 5 mm, added in 3-fold excess to recombinant HtsA and incubated at room temperature for ∼30 min. Protein solutions were then passed over a Sephadex G-50 (GE Healthcare) column and concentrated to 25 mg/ml.

Staphyloferrin A Synthesis and Purification

Using recombinant SfaB and SfaD, staphyloferrin A biosynthesis reactions were set up as previously described (33) and incubated for 12 h. The staphyloferrin A reaction was centrifuged in an Amicon® Ultra-0.5 10k filter column (Millipore) at 14,000 × g for 15 min to remove enzymes. The filtrate was then supplemented with 3 mm FeCl₃ and centrifuged at 18m000 × g to remove precipitate. 50 μl of the solution was then injected onto a Waters xTerra C18 reversed-phase 5-μm column (150 mm x 2.1 mm) on a Beckman System Gold HPLC equipped with a photodiode array detector. Samples were run at 0.2 ml/min using a step gradient as previously described (33). Solvent A was 10 mm tetrabutylammonium phosphate, pH 7.3, in HPLC grade water (Fisher), and solvent B was 100% acetonitrile (Fisher). Data were analyzed using the 32 Karat Software Version 8.0 system, and peaks were monitored at 340 nm. The peak corresponding to Fe-SA eluted at 17 min and was collected. Collected fractions were then vacuum-centrifuged to dryness and resuspended in deionized water, and the concentration of iron was determined using atomic absorption spectroscopy (see below) before use in fluorescence titration experiments with HtsA.

Determination of Ferric-Staphyloferrin A Concentration

Atomic absorption spectrometry was used to determine the concentration of iron in HPLC-purified Fe-SA samples. The concentration of iron was used to determine the concentration of staphyloferrin A by assuming a 1:1 molar ratio in the Fe-SA complex. Samples were diluted in 1 m nitric acid before being drawn by an SPS 5 sample preparation system (autosampler) into a Varian AA240 atomic absorption spectrometer. Absorbance was detected by a iron/manganese hallow cathode lamp, which emits at 248.3 nm specific for iron detection. Absorbance data were analyzed and compared with a linear calibration curve based on known iron standards in ppm. Iron standards were diluted in 1 m nitric acid from an atomic absorption spectrometer certified 1000 ppm ± 1% stock (Fisher). Calibration standards were separately analyzed first before iron-siderophore samples.

Fluorescence Spectroscopy

Fluorescence titration experiments were performed at room temperature using recombinant HtsA (15 nm) in 50 mm HEPES, pH 7.4, in a Fluorolog-3 spectrofluorometer (ISA Instruments). The excitation and emission slits were set at 2.1 and 6.3 nm, respectively. The excitation and emission wavelengths were set at 280 and 334 nm, respectively. Titration experiments were performed on two separate occasions, each time in triplicate, and the values reported are an average of all data sets. Dissociation constants (K_d) and relevant parameters were calculated by fitting the fluorescence titration data for Fe-SA (across a concentration range between 0.22 and 226 nm ligand) to a one-site binding model accounting for ligand depletion. Data were analyzed by nonlinear regression analysis using the solver tool add-in from Microsoft Excel software, as described previously (22, 23).

Multiple Sequence Alignments

A BLAST (38) search of the NCBI non-redundant protein data base identified many homologous proteins. The top 100 hits (E values <7 × 10⁻³¹) were exported from NCBI and filtered to remove sequences above 80% identity. The 34 remaining sequences were aligned using the program T-coffee (39, 40). The alignment was manually adjusted and visualized using Jalview (41).

Crystallization

HtsA was exposed to ferrated concentrated spent culture supernatant from S. aureus grown under iron restriction to form the HtsA-staphyloferrin A complex. Two different conditions yielded diffraction quality holoprotein crystals. Crystal form 1 was grown in hanging drop plates by microseeding with apoHtsA crystals grown as previously described (32). The crystals formed in 0.1 m HEPES, pH 7.0, 24% Jeffamine ED-2001. The crystals were frozen in the same buffer with 26% Jeffamine ED-2001 and 15% glycerol. Crystal form 2 grew in sitting drop plates containing a 1:1 ratio of protein sample to well solution (0.05 m zinc acetate, 20% polyethylene glycol 3350). The crystal was frozen in well solution supplemented with 20% ethylene glycol.

Structure Solution and Analysis

X-ray data for crystal form 1 were collected at the Stanford Synchrotron Radiation Lightsource on beamline 9-2 at 1.00 Å wavelength. Data were processed to 1.3 Å resolution using HKL2000 (42). The protein crystallized in the P2₁ space group with one molecule in the asymmetric unit. The siderophore-bound HtsA structure was solved by molecular replacement using MolRep (43) from the CCP4 program suite (44) with the previously described apoHtsA structure (PDB entry 3EIW) as the search model (32). The structure was edited using COOT (45) and refined with Refmac5 (46). Data collection and refinement statistics are shown in Table 1.

TABLE 1.

X-ray data collection and refinement statistics

	Fe-SA-HtsA open	Fe-SA-HtsA closed
Data collectiona
Resolution range (Å)	50-1.30 (1.35-1.30)	50-2.20 (2.32-2.20)
Space group	P21	P21
Unit cell dimensions (Å)	a = 44.56, b = 43.52,	a = 52.28, b = 148.60,
	c = 75.32, β = 100.6°	c = 52.27, β = 117.1°
Unique reflections	70,082	35,958
Completeness (%)	99.2 (97.7)	99.2 (99.2)
Average I/σI	19.6 (3.1)	10.8 (4.1)
Redundancy	3.5 (3.1)	4.0 (4.0)
Rmerge	0.056 (0.355)	0.085 (0.309)
Refinement
Rwork (Rfree)	15.3 (18.6)	16.5 (21.6)
B-factors (Å2)
All atoms	16.6	37.1
Protein	15.0	37.1
Staphyloferrin A	24.3	25.5
Water	28.0	38.1
r.m.s.d.b bond length (Å)	0.013	0.013
Ramachandran plot, % residues
In most-favorable region	92.5	86.7
In disallowed regions	0.0	0.0

^a Values in parentheses represent the highest resolution shell.

^b r.m.s.d., root mean square deviation.

X-ray data for crystal form 2 were collected at the Canadian Light Source on beam line 08ID-1 using a wavelength of 0.97934 Å. Data were processed to 2.2 Å using Mosflm (47) and Scala (48). Indexing suggested an apparent space group of C222₁. The structure was solved by molecular replacement, refined, and edited as described for crystal form 1. However, poor statistics, namely high values and large discrepancies between R_work and R_free values despite good density maps in addition to near identical values for a and c cell dimensions, suggested the space group was P2₁ (a = 52.28, b = 148.60, c = 52.27, β = 117.1°) with twinning. Several cases of twinning in P2₁ by the operator l, -k, h have been recently described (49–52). The revised solution in P2₁ contained two molecules in the asymmetric unit and was refined using Refmac with amplitude-based twin refinement (46). A twinning fraction of 0.49 (twinning operator l, -k, h) dramatically improved R_work (0.21 to 0.18) and R_free (0.29 to 0.24). The model was further refined with TLS parameters (53, 54) to a final R_work and R_free of 16.5 and 21.6, respectively.

Staphyloferrin A coordinates were generated using the program Sketcher from the CCP4 Program Suite (44). The central iron was identified in the electron density by a large peak in the difference map. Fe-SA was modeled into the structures after fitting the protein backbone before adding waters. Fe-SA was modeled into the closed structure at full occupancy and has an average B-factor of 28.5 Ų. The open conformation was modeled with Fe-SA at 0.70 occupancy, as determined to minimize density peaks in an F_o − F_c map at the iron atom. Fe-SA refined with an average B-factor of 34.6 Ų. Data collection and final refinement statistics are shown in Table 1. Figures were generated with PyMol (55).

RESULTS

Affinity of HtsA for Staphyloferrin A

Our previous studies showed that HtsABC was required for SA utilization in S. aureus (32). HtsA, a class III substrate binding protein, is tethered to the extracellular face of the cytoplasmic membrane via an N-terminal lipidation and functions as the receptor for staphyloferrin A. Changes in the intrinsic fluorescence of recombinant HtsA (lacking signal peptide) that occur upon ligand binding were examined to determine substrate affinity. Saturating concentrations of HPLC-purified ferric-SA resulted in an average 52.5% reduction in fluorescence emission. The dissociation constant (K_d) of HtsA and Fe-SA was in the low nm range but could not be accurately determined because the concentration of HtsA required to see fluorescence change (15 nm) was ∼15-fold greater than the K_D (supplemental Fig. S1). The specificity of HtsA for Fe-SA was demonstrated by the fact that ferric-staphyloferrin B (staphyloferrin B was synthesized as described (32) and HPLC-purified in a similar fashion to SA) failed to quench the intrinsic fluorescence of HtsA. These data are in agreement with previously published biological data (31, 32).

Overall Structure of Open and Closed HtsA

HtsA possesses distinct N- and C-terminal domains joined by a single α-helix (56, 57) (Fig. 1A). The apoHtsA structure (PDB entry 3EIW (32)) overlays with the Fe-SA-bound crystal form 1 (open) and crystal form 2 (closed) HtsA structures with an root mean square deviation over all Cα atoms of 0.5 and 1.6 Å, respectively (Fig. 1B). Unlike the typical interdomain movement characteristic of many β-sheet bridged binding proteins, the N- and C-terminal domain cores overlay well with the apo structure. Relative to the apo structure, the open holo structure does not undergo significant domain movement (hinged motion of less than 2°) Instead, the predominant structural difference is centered at three loops at the surface of the C-terminal domain. These three loops are composed of residues 201–208 (Loop^201–208), 228–258 (Loop^228–258), and 265–271 (Loop^265–271) (Fig. 1B). In the open structures, residues in the loops display elevated B-factors relative to core residues (Fig. 1C), whereas in the closed structure, the B-factors are more similar (Fig. 1D). Loop^228–258 undergoes the largest structural change in the closed protein with Cα atoms moving as much as 12.1 Å (Tyr-239) across the binding pocket relative to the apo or open structures. The large loop movement is accommodated by a slight unwinding of the α-helix^230–239 preceding the loop. Tyr-239 is located at the C-terminal end of the α-helix^230–239 and is translated from the central portion of the C-terminal domain into the binding pocket to form an H-bond to the lateral side of Fe-SA. A second Tyr, Tyr-244, is shifted across the pocket, forming a H-bond (2.5 Å) to the Phe-146 main chain carbonyl of the N-terminal domain. The loop movement also creates several intradomain H-bonds. Lys-238 forms a H-bond to Tyr-212 (3.0 Å). Hydrophobic contacts are also created by the loop movement. Leu-240 forms a hydrophobic contact with Phe-146, again bridging the N- and C-terminal domains. Pro-243 forms a stacked hydrophobic interaction with Tyr-244 (3.8 Å), potentially stabilizing the interdomain contacts facilitated by Tyr-244.

FIGURE 1.

The overall structure of the HtsA-staphyloferrin A complex. A, the open structure (crystal form 1) of HtsA is shown as a schematic colored in a gradual color change from the N terminus (blue) to the C terminus (red). Staphyloferrin A is shown in the binding pocket as sticks with carbon, nitrogen, oxygen, and iron shown in gray, blue, red, and orange, respectively. B, shown is the overlay of open holoHtsA (blue, crystal form 1) and closed holoHtsA (red, crystal form 2) and apoHtsA (green, PDB entry 3EIW). The structure is rotated ∼90° relative to A to look into the binding pocket. Backbones are shown as tubes. Staphyloferrin A is shown in the binding pocket as in A. C, shown is a B-factor tube diagram of open-HtsA. Regions of increasing B-factor are shown with larger diameter and coloring from blue (low) to red (high). D, shown is a B-factor tube diagram of closed-HtsA. The structure is colored as in C.

In the closed structure, two Zn²⁺-mediated crystal contacts are formed between symmetry related C-terminal domains of each molecule. One site involves two metals bound by His-266 and Lys-270 from chain A and Glu-250, His-251, and Asp-254 from a symmetry-related chain B. The other site involves the same five residues, this time with the Glu-250, His-251, and Asp-254 from Chain A and His-266 and Lys-270 from a symmetry-related chain B. Zn²⁺ is present in the crystallization buffer and has been modeled into both sites. Because these crystal contacts involve residues from the loops with the largest structural changes, it is feasible that these crystal contacts induce a conformational change in holo-HtsA. Given the additional Fe-SA and interdomain HtsA contacts formed, a more likely explanation is that the closed conformer seen is biologically relevant and the crystal contacts simply form between two stable, closed structures.

Structure of Staphyloferrin A

SA is synthesized from two molecules of citrate forming amide bonds with the amino groups of a d-ornithine, forming N²,N⁵-di-(l-oxo-3-hydroxy-3,4-dicarboxylbutyl)-d-ornithine (Fig. 2A) (28). Electron density for Fe-SA is present in the binding pocket formed between the N- and C-terminal domains in both the open and closed forms of HtsA. Electron density is present for the complete Fe-SA molecule in the closed structure, but weak density for the ornithine backbone and additional density around one terminal carboxylate group is apparent in the open structure (Fig. 2B). Two waters have been modeled into the extra density but do not completely account for the extra positive peaks. The ligand is likely afforded additional flexibility in the absence of interaction by Loop^228–258 that moves across the pocket in the closed structure.

FIGURE 2.

The structure and chirality of staphyloferrin A. A, a linear schematic of the staphyloferrin A molecule is shown. Stereochemistry at the three chiral centers is indicated. Atoms that directly interact with the iron are numbered according to Konetschny-Rapp et al. (28). B, shown is the conformation of staphyloferrin A from the open HtsA-SA structure. Extra density can be seen at the distal end of the terminal carboxylate group (group 3 from 2A). C, shown is the conformation of SA from the closed HtsA-SA structure. The omit F_o − F_c maps are contoured at 2.5 and 3.5 σ for B and C, respectively. Atoms are indicated by their element symbols.

The SA structure has the ornithine Cα atom in the R-configuration. The two chiral centers of the citrate components are modeled into the density for both structures with the more buried citrate in the S configuration and the more solvent-exposed citrate in the R configuration (Figs. 2, B and C). The citrates on either side of the iron bind as mirror images of one another. The chirality at the citrate carbons are in line with predictions from an early characterization of SA, suggesting the chiral centers were likely R,S (28). However, subsequent findings by a different group based on the CD spectra of model compounds suggested the chiral centers were S,S (26). Modeling of SA with the S,S configuration does not fit the observed electron density as well in either the open or closed structures. Modeling in the second S configuration clearly strains the siderophore providing poor fit to the density for O1, O2, and O3 groups (see supplemental Fig. S2 for a comparison of S,S and S,R models).

Ferric iron is coordinated by an oxygen atom from each of β-hydroxy,β-carboxyl-substituted carboxylates from the two citrate constituents. SA coordinates Fe³⁺ with distorted octahedral geometry and ligand bond lengths of 2.0–2.2 Å. The ligand bond angles are distorted from perfect octahedral geometry ranging from 75° to 101° (Fig. 2, B and C).

Siderophore Bound in the Open Conformation of HtsA

The basic patch previously suggested as the putative Fe-SA binding site contributes the majority of siderophore contacts (32). In the open structure, five Arg residues at the pocket surface form direct H-bonds to oxygen atoms of Fe-SA (Fig. 3A and Table 2). Arg-104 and Arg-126 form H-bonds to Fe-SA. Arg-299 is modeled in two conformations, both within H-bonding distance of the Fe-SA ornithine hydroxyl group. Arg-304 and Arg-306 forms an additional H-bond to the terminal carboxylate group of the more buried terminus. His-209 forms an additional H-bond to the hydroxyl of the ornithine component, and four ordered water molecules are modeled interacting with the ornithine carboxylate and the carbonyl as well as two carboxylates from one of the citrate moieties (Fig. 3A and Table 2). The B-factors of regions of the siderophore backbone are dependent on solvent accessibility. The more buried citrate group has the lowest B-factors (carbon atoms ranging from 14–18 Ų). The carbon atom B-factors of the ornithine increase from 20 to 36 Ų, moving from the inner citrate group to the fully solvent-exposed ornithine carboxylate. The outermost citrate group carbons have B-factors in the 24–30 Ų range.

FIGURE 3.

Staphyloferrin A in the HtsA binding pocket. A, shown is the open HtsA binding pocket. Residues forming direct contacts with SA (cyan) are shown as sticks with carbon, nitrogen, oxygen, and iron shown as green, blue, red, and orange, respectively. Hydrogen bonds are indicated by dashed lines. Residues are numbered according to full-length HtsA. B, the closed HtsA binding pocket shown colored as in A.

TABLE 2.

Fe-SA-HtsA bond distances (Å)

HtsA atom-Fe-SA atoma	Bond distance
	Fe-SA-HtsA open	Fe-SA-HtsA closed
	Å
Arg-86 NΗ-O2	NAb	3.1
Arg-104 NΗ1, NΗ2-O3′, O3′	2.9, 3.0	2.8, 3.0
Arg-126 NΗ-O3′	3.5	3.5
Arg-126 NΗ1, NΗ2-O3, O3	2.8, 3.1	2.7, 3.1
Lys-203 O-ornithine carboxylate	NAb	2.9
His-209 Nϵ2-citrate′ carbonyl	2.8	2.7
Tyr-239 OΗ-ornithine N, O1	NAb	2.8, 2.6
Arg-299 Nϵ-O2′	3.0 or 2.7c	NAb
Arg-299 NΗ-O2′, O2′	2.8, 3.6 or 3.1, 3.3c	2.7, 3.1
Arg-299 NΗ-citrate carbonyl	2.9 or NAb,c	NAb
Arg-304 Nϵ-O2′	3.4	3.4
Arg-304 Nϵ-O2	3.3	3.1
Arg-304 NH-O2	3.2	3.5
Arg-306 NΗ-O3′	2.8	2.4
Water-ornithine carboxylate	2.9	NAb
Water-citrate carbonyl	3.0	NAb
Water-O2	3.1	NAb
Water-O3	3.3	3.0

^a Fe-SA atoms are numbered according to atom labels in Fig. 2A.

^b NA, not applicable because residues contact not present in structure.

^c Refers to each of the two conformations modeled for Arg-299 in open structure.

Siderophore Bound in the Closed Conformation of HtsA

The major difference in HtsA-Fe-SA interactions between the open and closed conformations occur due to additional contacts from two of the loops undergoing large conformational changes. Relative to the open conformation, Tyr-239 Cα translates 12.1 Å in Loop^228–258 and forms a H-bond with the carbonyl group of the less buried citrate. Additionally, the translation of Loop^201–208 orients Lys-203 and Arg-83 to form H-bonds with the Fe-SA carboxylate groups (Fig. 3B and Table 2). Despite these additional contacts, Fe-SA is bound within the binding pocket of the closed HtsA in a similar orientation to the open form; however, it is shifted slightly deeper into the pocket (Fig. 3B). The result of the conformational changes is significant occlusion of solvent from the Fe-SA molecule, reducing the 33.0% solvent exposure in the open structure to 14.5% in the closed structure (as determined with AREAIMOL (44)). The distribution of siderophore backbone B-factors is similar to that of the open conformation. The lowest B-factors are associated with the more buried citrate group (15–20 Ų), a broad range is observed in the ornithine group (25–40 Ų), and the outer citrate group has elevated B-factors (27–30 Ų).

In an overlay of the open and closed protein structures there is an ∼1.5 Å average atom displacement over all 34 Fe-SA atoms. The half of Fe-SA located closest to the exterior of the HtsA binding pocket undergoes the largest displacement with an average shift of ∼2.1 Å into the pocket. Only slight intramolecular atomic displacements occur within Fe-SA with mean atom displacements relative to the protein core of ∼0.5 Å (Fig. 2B).

Several of the additional key protein-Fe-SA contacts identified in the open structure are similar to those in the closed structure; however, H-bond-lengths are altered in many cases (Fig. 3B and Table 2). In total, six Arg residues in the Arg-rich region form direct contacts with Fe-SA. Arg-86, Arg-104, Arg-126, Arg-299, Arg-304, and Arg-306 all form H-bonds to O atoms in the citrate moieties of Fe-SA (Fig. 3B and Table 2). Also similar to the open structure, His-209 forms a H-bond to the carbonyl group on the ornithine component of Fe-SA.

Multiple Sequence Alignments

A BLAST (38) search of the NCBI non-redundant protein data base identified proteins with greater than 30% sequence identity. The list of sequences is of proteins from Gram-positive and Gram-negative bacteria. Because several staphylococcal species produce SA (28, 29), the best matches correspond to homologous receptor proteins in Staphylococcus species aureus, epidermidis, warneri, haemolyticus, capitis, saprophyticus, and hominus with amino acid sequence identities of >80%. Three proteins were identified from Bacillus sp.: (i) YfmC (∼42% identity), annotated as a ferric citrate-binding protein (58), (ii) YhfQ (∼37% identity), an unknown siderophore-binding protein (58, 59), and (iii) YfiY (∼30% identity), the binding protein for the siderophore schizokinen (60). Another interesting homolog identified was the ferric citrate-binding protein, FecB, found in many organisms but probably best studied in E. coli (E. coli FecB, ∼35% identity).

An alignment of 41 of the homologous sequences was made (for a subset of sequences, see Fig. 4). The HtsA sequence contains an ∼8-residue insertion relative to most homologous sequences in the alignment around Tyr-239, in Loop^228–258, which undergoes the largest structural change and forms a direct Fe-SA contact in the closed structure (Figs. 1B and 4). This suggests the structural change will not be seen in many homologous structures and may be an adaptation for entrapment of specific substrates.

FIGURE 4.

HtsA sequence alignments. A subset of the 34 non-redundant sequences (<80% sequence identity) is shown aligned. Residues are numbered as in full-length HtsA. Sequences are identified by the protein name or identifier followed by an underscore, and the first uppercase letter of the genus name followed by the first three letters of the species name. Arrows indicate HtsA residues that directly interact with SA.

The residues contacting Fe-SA are conserved to varying degrees (Arg-86 (22/41), Arg-104 (37/41), Arg-126 (36/41), His-209 (30/41), Tyr-239 (9/41), Arg-299 (21/41), Arg-304 (20/34, 6 Lys), Arg-306 (37/41)), although in many cases the amino acid substitutions are conservative. Furthermore, two conserved Glu residues, Glu-110 (39/41, 2 Asp) and Glu-250 (38/41, 2 Asp, 1 Ser), are located such that they could form salt bridges between the N- and C-terminal domains of HtsA and conserved Arg residues on the ABC transporter permease components (HtsBC) to mediate protein docking. Interestingly, Glu-250 is located on an α-helix that shifts ∼2.8 Å toward the domain interface in the closed structure. A model of the HtsBC permease based on the BtuF structure (PDB entry 2qi9 (61)) suggests the movement of Glu-250 brings it toward a conserved Arg-74 of HtsB or Arg56 of HtsC, which may mediate differentiation of ligand-bound or free HtsA.

Several functionally interesting residues are highly conserved (Fig. 4). Gly-102 (41/41) and Pro-107 (39/41) occur on either side of the conserved Fe-SA-interacting Arg-104, orienting the arginine into the binding pocket. His-127 (40/41) forms a H-bond to the main chain O of Arg-126 and likely stabilizes the loop containing SA-interacting Arg-126. Four sequential residues, Ile-137—Thr-140, are conserved in most sequences and occur on a tightly turning surface loop. Tyr-150 (35/41) forms a H-bond between the N-terminal domain and the His-177 (33/41) Nϵ from the domain-spanning α-helix, which would contribute to interdomain stability. Trp-302 (41/41) is located close to two Fe-SA-ligands (Arg-304 and Arg-306) but is directed into the core of the protein where it forms several hydrophobic contacts in addition to a H-bond from its Nϵ1 group to Glu-317. Trp-302 burial likely imparts stability to the loop and anchors the ligand binding Arg residues. Similar to the N-terminal linkage, a domain linking interaction between the C-terminal domain and the bridging α-helix is mediated by a H-bond formed between Glu-312 (32/41) and Arg-173 in both the open and closed conformations. However, in the closed conformation, Arg-173 models in two orientations, both bonding and non-bonding.

DISCUSSION

The crystal structures of Fe-SA-bound HtsA in the open and closed forms have enabled identification of residues that interact with Fe-SA and demonstrated an unprecedented mode of ligand entrapment not previously observed in this family of binding proteins. Several x-ray crystal structures of related class III binding proteins possessing similar structural folds have been determined in both apo and holo forms. BtuF (E. coli B₁₂ uptake) (56, 62), TroA (E. coli Zn²⁺ uptake) (63, 64), FhuD (E. coli ferrichrome uptake) (65, 66), and ShuT (Shigella dysenteriae heme uptake) (67) all display little to no conformational change between the apo and holo forms (<4° hinge movement and small intradomain atom displacement).

A few recent examples demonstrate that a larger interdomain movement is possible. The apo structure of E. coli FitE was recently presented with both open and closed conformations found within the four molecules of the asymmetric unit. Within each domain FitE undergoes little conformational change (0.4–0.75 Å root mean square deviation for Cα atoms). However, the domain-bridging α-helix undergoes a significant hinged motion (∼18.5°) (68). The structure of the Bacillus subtilis bacillibactin receptor (FeuA) undergoes a hinged movement of ∼20° between the apo (PDB entry 2phz) and holo structures (69). Consistent with these crystal structures, molecular dynamics simulations of FhuD predict a greater hinge motion (∼6°) than what is observed between the apo and holo structures (2°) (66).

HtsA undergoes significant conformational changes upon Fe-SA binding. However, the conformational changes do not mirror the rigid interdomain movement seen in the FitE or FeuA structures or FhuD simulations. Instead, the large conformational changes are isolated to specific regions within the C-terminal domain. These conformational changes allow HtsA to clamp around Fe-SA, providing additional contacts to the siderophore as well as interdomain contacts that may facilitate the slight hinge closing motion. Similar isolated conformational changes of this magnitude have not previously been observed upon ligand binding in class III binding proteins.

The ligand-dependent conformational changes in class III binding proteins likely affect docking to the permease component of the ABC transporter, thereby providing a means of discriminating between the ligand-bound versus ligand-free receptor. The BtuCD-F complex crystal structure has been determined, demonstrating that the binding of BtuF to BtuC is mediated by salt bridges between Glu-72 and Glu-202 from BtuF and Arg-56 and Arg-295 from BtuB (61). Alignments and subsequent site-directed mutagenesis of two similar class III binding proteins, E. coli FecB (70) and S. aureus FhuD2 (22), suggest that similar Glu-Arg salt bridges mediate docking as variants affect ligand transport but not ligand binding. Because docking is mediated by salt bridge formation, even minimal hinged motion would allow discrimination of open and closed conformations. The closed HtsA structure illustrates an alternative mechanism of discrimination of the ligand-bound form of the receptor. The movement of Loop^228–258 to enclose Fe-SA alters the placement of the homologous Glu-250, which likely mediates salt-bridge formation with the permease HtsBC.

The HtsA-Fe-SA complex is the first complex of an α-hydroxycarboxylate-type siderophore bound to its cognate binding protein, and to our knowledge this study reports the first examination of affinity between an α-hydroxycarboxylate siderophore and its receptor. The K_d of HtsA for Fe-SA, in the low nm range, provides the explanation for how HtsA, when exposed to ferrated S. aureus culture supernatant, was able to complex and crystallize with Fe-SA. The combination of a large number of charge interactions and receptor closures around the siderophore afford specificity to the interaction of SA with HtsA. Many siderophore-binding proteins and transporters have broad specificity for siderophores of a similar type; however, the Hts system is highly specific. SB is also an α-hydroxycarboxylate-type siderophore (30), yet the Hts system cannot transport sufficient amounts for growth (32). The only common constituent is a single citric acid component. Staphyloferrin B is also composed of a l-2,3-diaminopropionic acid, 1,2-diaminoethane, and a succinic semialdehyde (30). The specificity of the Hts system for one negatively charged α-hydroxycarboxylate siderophore over another is likely a reflection of the specific ionic contacts formed in the closed structure that would not properly accommodate staphyloferrin B. Although the advantage gained by this specificity is unclear, it could be tailored to the low concentrations the siderophores expected to be present in serum at the point of infection.

The affinity of HtsA for Fe-SA is within the range of several outer membrane siderophore receptors in Gram-negative bacteria. Compared with other class III substrate binding proteins, the binding affinity of HtsA for SA is greater than the presented Bacillus sp. receptors and orders of magnitude stronger than E. coli FhuD (Table 3). These apparent differences in affinity may be a reflection of the extent and type of residues involved in siderophore binding or may result from differences in conformational changes or tryptophan masking that can affect fluorescence emission. The disparity between binding affinities for the E. coli hydroxamate-binding protein, FhuD, and HtsA could reflect the differences in specificity. FhuD is a receptor for a broad array of ferric hydroxamates, so the binding pocket sacrifices affinity for diversity, where HtsA is specialized for SA alone.

TABLE 3.

Dissociation constants (K_d) for receptor-ferric-siderophore complexes

Protein-siderophore (organism)a	Kd	Reference
	nm
HtsA-staphyloferrin A (S. aureus)	Low nm	This Study
FeuA-bacillibactin (B. cereus)	19	60
FeuA-enterobactin (B. cereus)	12	60
FatB-3,4-DHB (B. cereus)	1.2	60
FatB-petrobactin (B. cereus)	127	60
FpuA-petrobactin (B. cereus)	175	60
YfiY-schizokinen (B. cereus)	34	60
YxeB-desferroximine (B. cereus)	18	60
YclQ-petrobactin (B. subtilis)	113	78
FhuD-ferric hydroxamates (E. coli)	300–7900	82
FpvA*-pyoverdin (P. aeruginosa)	0.37	83
FhuA*-ferrichrome (P. aeruginosa)	0.65	83
FptA*-pyochelin (P. aeruginosa)	0.54	83

^a Proteins listed are class III ligand-binding proteins, except for those with asterisks, which are outer membrane receptor proteins in Gram-negative bacteria.

The chemical characteristics of siderophore binding pockets of receptors varies with the class of the cognate siderophore. Ferric complexes of hydroxamate-type siderophores are generally hydrophobic. Crystal structures of the outer membrane receptor FhuA (71) and the binding protein FhuD (65, 72) from the E. coli hydroxamate uptake system have been determined. Siderophore binding in both structures is primarily mediated through hydrophobic contacts. Similarly, in Pseudomonas aeruginosa, receptors for the largely hydrophobic siderophore ferric chelates, pyoverdin (FpvA) (73) and pyochelin (FptA), have binding sites that are primarily composed of hydrophobic and aromatic residues (74).

Catecholate-type siderophore ferric chelates generally have a net negative charge, and the receptor binding residues often mirror the net charge. Ferric enterobactin carries a −3 net charge, so not surprisingly, the binding pockets of the E. coli outer membrane receptor, FepA (75), and the Campylobacter jejuni-binding protein, CeuE (76), contain several positively charged residues. In CeuE, the net negative charge is balanced by three Arg residues, and although the binding residues in FepA could not be definitively identified, an Arg-rich binding site was found (75, 76). Interestingly, a hydrophobic patch was identified in FepA and shown by mutagenesis data to contribute to the affinity for ferric enterobactin (77). The structures of the catecholate receptors FeuA (Bacillus cereus) and YclQ (B. subtilis), which bind ferric-bacillibactin and ferric-petrobactin, respectively, represent the only other Gram-positive bacterial siderophore receptor structures determined to date. Recently, a FeuA-ferric-bacillibactin crystal structure was determined (69). Ferric-bacillibactin carries a net negative charge that is neutralized by three basic residues, Lys-84, Lys-105, and Arg-180, that interact with deprotonated catecholate oxygen atoms (69). Gln-181 and Gln-215 also form direct H-bonds to bacillibactin. Furthermore, the YclQ ligand binding cleft contains a series of conserved positively charged residues (Arg-104, Arg-192, Arg-236, and Try-275) that are predicted to interact with a negatively charged petrobactin ligand (78).

Crystal structures of mammalian siderocalin bound to ferric complexes of catecholate-type siderophores have been determined (79, 80). The crystal structures show that two Lys residues and a single Arg mediate binding to the negatively charged catecholates. Recently, it was established that electrostatic interactions between ferric siderophores and siderocalin are the prime determinants of binding affinity (81).

Analogous to these examples of charged siderophores, Fe-SA carries a net negative charge that is neutralized by six Arg residues in HtsA. This large number of positively charged residues is expected to favor tight binding to a compound that would be present in low concentrations in the environment. The abundance of positive electrostatic potential, in concert with the occlusion of the binding site upon closing, likely determines the specificity of the transporter for Fe-SA. Furthermore, the extensive protein-siderophore contacts that enclose the small siderophore likely serve to discriminate between Fe-free and Fe-bound SA.

In summary, we have demonstrated that the S. aureus ABC transporter-associated binding protein HtsA binds Fe-SA and undergoes conformational changes upon binding involving a very small scale hinge motion and relatively large movements at loops in the C-terminal domain to enclose the ligand. Furthermore, binding is mediated primarily by six Arg, a Tyr, and a His residue in the binding pocket. The coordinating residues are well conserved in several proteins, suggesting that a similar means of coordination may be utilized in both siderophore and exogenous ferric citrate uptake pathways from a broad range of bacteria.