Identification and Characterization of a Novel-type Ferric Siderophore Reductase from a Gram-positive Extremophile

Abstract

Iron limitation is one major constraint of microbial life, and a plethora of microbes use siderophores for high affinity iron acquisition. Because specific enzymes for reductive iron release in Gram-positives are not known, we searched Firmicute genomes and found a novel association pattern of putative ferric siderophore reductases and uptake genes. The reductase from the schizokinen-producing alkaliphile Bacillus halodurans was found to cluster with a ferric citrate-hydroxamate uptake system and to catalyze iron release efficiently from Fe[III]-dicitrate, Fe[III]-schizokinen, Fe[III]-aerobactin, and ferrichrome. The gene was hence named fchR for ferric citrate and hydroxamate reductase. The tightly bound [2Fe-2S] cofactor of FchR was identified by UV-visible, EPR, CD spectroscopy, and mass spectrometry. Iron release kinetics were determined with several substrates by using ferredoxin as electron donor. Catalytic efficiencies were strongly enhanced in the presence of an iron-sulfur scaffold protein scavenging the released ferrous iron. Competitive inhibition of FchR was observed with Ga(III)-charged siderophores with K_i values in the micromolar range. The principal catalytic mechanism was found to couple increasing K_m and K_D values of substrate binding with increasing k_cat values, resulting in high catalytic efficiencies over a wide redox range. Physiologically, a chromosomal fchR deletion led to strongly impaired growth during iron limitation even in the presence of ferric siderophores. Inductively coupled plasma-MS analysis of ΔfchR revealed intracellular iron accumulation, indicating that the ferric substrates were not efficiently metabolized. We further show that FchR can be efficiently inhibited by redox-inert siderophore mimics in vivo, suggesting that substrate-specific ferric siderophore reductases may present future targets for microbial pathogen control.

Introduction

Siderophore-dependent iron acquisition is an essential metabolic feature employed by a vast number of bacteria, fungi, plants, and even higher eukaryotes (1–4). Key steps of typical siderophore pathways include siderophore synthesis, secretion, and uptake of siderophore-bound iron that is coupled to its intracellular or extracellular release. Within siderophore pathways, iron release processes are still widely uncharacterized. Generally, two enzymatic strategies are known for direct release of ferric siderophore complexes, which are hydrolysis of the siderophore backbone or reduction of the complexed ferric iron species, thus representing either a scaffold- or metal-targeted release mechanism (5–7), which may not necessarily be mutually exclusive. Iron release outside the cytosol, including compartments such as the bacterial periplasm or eukaryotic vacuoles, may further be coupled to protonation of ferric siderophore complexes (8, 9).

Hydrolytic release of iron is restricted to a small number of siderophores possessing bonds that can be efficiently attacked by water. Usually, these are ester bonds that are present in trilactone siderophores like enterobactin, bacillibactin, salmochelins, or triacetylfusarinine C. For those siderophores, several esterases have been described that partially or completely hydrolyze these intramolecular ester bridges (10–14).

The overwhelming majority of siderophores is assembled by amide bond formation and thus are very robust against hydrolysis. Their ferric complexes are generally released by metal reduction and/or complex protonation. Reductive release in the extracellular environment has been described for the membrane-standing ferric reductases in yeast, especially Fre1p, Fre2p, Fre3p, and Fre4p (15). They are similar to b-type cytochromes and belong to the flavocytochrome superfamily using FAD, NAD(P)H, and heme cofactors for electron transfer during catalysis. Electron shuttling across the cytosolic membrane is suggested to be coupled with proton transfer resulting in extracellular acidification (16), which may increase the redox potentials of the ferric complexes. Generally, the iron-chelate redox potentials greatly differ among the different classes of siderophores. Triscatecholates such as ferric enterobactin (Fe(III)-enterobactin) with iron binding affinities in the range of 10⁴⁹ m⁻¹ possess standard redox potentials of their ferric complexes at E′₀, pH 7.0, of −750 mV or lower (17, 18). Redox potentials of hydroxamates such as ferrichrome or ferrioxamines, and citrate-hydroxamates such as ferric aerobactin (Fe(III)-aerobactin) are higher but still in the negative range below an E′₀, pH 7.0, of about −300 mV, whereas carboxylates such as ferric dicitrate (Fe(III)-dicitrate) are ranging above them at E′₀, pH 7.0, of about 0 mV (6). According to the physiological range of cellular redox compounds, many ferric siderophores can be potentially reduced by soluble redox cofactors, a mechanism that has mainly been described for cytosolic iron release in bacteria (6, 18). In these cases, usually flavin reductases catalyze electron transfer from NAD(P)H toward different flavins such as FMN, FAD, or riboflavin. These flavins may then be released from the enzyme as free reducing agents such as in Escherichia coli NAD(P)H:flavin oxidoreductase Fre and its homologs in Vibrio (19), the sulfite reductase SiR (20), and Magnetospirillum gryphiswaldense flavin reductase FeR (21) or may stay enzyme-bound such as in E. coli flavohemoglobin Hmp (22), nitroreductase NfnB, and ferredoxin-NADP⁺ reductase Fpr (23). In addition, extracellular FSRs² dependent on NADH and flavins were described in several species, including E. coli, Yersinia enterocolitica, Pseudomonas aeruginosa, and Listeria monocytogenes (24). Generally, these FSRs act on a broad set of substrates, including iron-loaded non-siderophores such as cytochrome c or ferredoxin and non-metals such as sulfite or nitro compounds, and hence are not specifically linked with iron assimilatory metabolism and regulation. Still, there are few examples of cytosolic FSRs showing a direct relation with iron metabolism as well as direct interaction with a defined set of ferric chelate substrates. One reductase that is suggested to fulfill these criteria is E. coli FhuF, which was shown to reduce several ferric hydroxamate complexes (25, 26). FhuF contains a C-terminal iron-sulfur (Fe/S) cluster that likely permits electron transfer from the enzyme to its cognate substrates. Although FhuF-type reductases are present in several enterobacterial species, further types of putative reductases with C-terminally conserved cysteine motif are distributed in other phyla including Firmicutes. Their reductases have not been characterized yet, and it has not been described that Gram-positive bacteria possess iron-regulated or substrate-defined FSRs.

Here, we describe for the first time a distinct FSR from a Gram-positive bacterium. The reductase was found to be iron-regulated and to possess a stably bound low-potential Fe/S cofactor. Its kinetic analysis revealed a redox-scaled substrate spectrum defined by both cluster midpoint potential and substrate affinity. Genetic analyses showed that its function is decisive for cellular iron metabolism and that it can be inhibited by redox-inert siderophores during iron deprivation. We thus present novel mechanistics insights into cytosolic iron release, which might have implications for antibiotic strategies.

EXPERIMENTAL PROCEDURES

Growth Conditions, DNA Preparation, and Cloning

Bacillus halodurans DSM497 (DSMZ stock, Braunschweig, Germany) was routinely cultured in LB medium containing 0.1 mm sodium sesquicarbonate (4.2 g of NaHCO₃, 5.3 g of Na₂CO₃ per liter) at pH 9.7 (“LB-ha broth”) at 37 °C. DNA was prepared from late log phase cultures by the following procedure. After washing cells twice with TE buffer (10 mm Tris-HCl, pH 8.0, 1 mm Na-EDTA), 0.1 mm lysozyme was added, and the solution was incubated for 10 min at 37 °C. Upon addition of 2% (w/v) SDS and 1 m sodium perchlorate, cell proteins were precipitated, and after extraction with chloroform/isoamyl alcohol at 24:1 (v/v), DNA was separated into the aqueous phase. DNA was precipitated by addition of 2 volumes of ethanol, washed in 70% (v/v) ethanol, dried under vacuum, and dissolved in TE buffer. RNA was removed by RNase I treatment (0.1 mg/ml) for 1 h at 37 °C. For amplification of genes BH1040 and BH1037, primer pairs ATATCTAGATAACGAGGGCAAAAAATGATCGAGCCACCTGTTATGAATG (BH1040_forward)/GCTCGGTTTGCGAGGGCACGTCC (BH1040_reverse), and ATAGGATCCGGAGGAAATGAACCAAGCGAAG (BH1037_forward)/ATAAAGCTTCTATTGTGTAAGGGAATCAACGAG (BH1037_reverse) were used, respectively, according to the data base genome sequence of B. halodurans C-125 (restriction sites are underlined, primer BH1040_reverse blunt end and phosphorylated). BH1040 and BH1037 were cloned into pMM30 and pCB28a+ (12) resulting in pMM30-1 and pMM20 expression vectors providing a C-terminal Strep-tag II and an N-terminal His₆-tag, respectively. In vitro DNA manipulations and E. coli transformations were done according to described methods (27).

Recombinant Protein Production and Purification

E. coli BL21 cells transformed with the desired expression vector were grown in LB medium at 37 °C until an A₆₀₀ of 0.5 and then induced with 0.2 mg/liter anhydrotetracycline (pMM30-1) or 0.2 mm isopropyl β-d-thiogalactopyranoside (pMM20) for 3 h at 30 °C. For purification of recombinant B. halodurans BH1040, cells were harvested, resuspended in 150 mm NaCl, 100 mm Tris-HCl, pH 8.0, and disrupted by using a French press (Thermo Scientific), and the filtrated lysate was subjected to Strep-Tactin chromatography using an FPLC purifier system (GE Healthcare) and a column with 2 ml of Strep-Tactin superflow material (IBA). The recombinant protein was eluted with 2.5 mm d-desthiobiotin (IBA). Purification of recombinant B. halodurans BH1037 and recombinant Bacillus subtilis SufU was done by Ni²⁺-nitrilotriacetic acid chromatography as described previously (12, 35). The pooled and concentrated elution fractions were then applied to size exclusion chromatography using a 26/60 Superdex 200 column and 150 mm NaCl, 100 mm Tris-HCl, pH 8.0. Fractions of the dominant single UV peak were analyzed by SDS-PAGE, and those containing pure protein were concentrated using centrifugal filter devices with a 10,000 cutoff. Protein concentration was determined by Bradford method (28) using a BSA calibration curve. Protein was shock-frozen in liquid nitrogen and stored at −80 °C.

UV-Visible Analysis

For measurement of oxidized protein spectra, purified protein was diluted to the desired concentration in 150 mm NaCl, 100 mm Tris-HCl, pH 8.0, and the solution was placed into a quartz cuvette (n = 1 cm). For measurement of reduced protein spectra, the protein was treated with 5 mm sodium dithionite under anaerobic conditions (95% N₂, 5% H₂). Reduced protein was placed into an anaerobic cuvette, which was tightly sealed before measurements were performed. Spectra were recorded at an Ultrospec 3000 spectrophotometer usually from 250 to 800 nm. Data were analyzed with SWIFT Scan 2.06.

Analytical Gel Filtration

A Zorbax GF-250 column was equilibrated with 150 mm NaCl, 100 mm Tris-HCl, pH 8.0, using an Agilent HPLC system (1200 series). The column was calibrated using a mixture of ferritin (440 kDa), aldolase (158 kDa), conalbumin (75 kDa), ovalbumin (44 kDa), carbonic anhydrase (29 kDa), and aprotinin (6.5 kDa) at 0.5 ml/min at 25 °C. 100 μg of purified recombinant FchR was then applied at the same flow rate and temperature. UV spectra at 280 nm were recorded and analyzed with Agilent ChemStation software B.03.

Western Analysis

B. halodurans wild-type (WT) and B. halodurans ΔfchR were cultivated in LB-ha broth supplemented with 100 μm FeCl₃ or different concentrations (5–50 μm) of 2,2′-bipyridyl (BP) and 1,10-phenanthroline (Phen). Cells were harvested at late log phase, washed twice with TE buffer, and disrupted after addition of 1 mm serine protease inhibitor PMSF by sonication (four times for 1 min at 50 watts). Cell debris was removed by centrifugation (15,000 × g, 60 min, 4 °C), and clear lysate was separated. Protein concentrations of cytosolic extracts were determined by the Bradford method, and 20 μg were applied to each lane for SDS-PAGE (12% acrylamide). Furthermore, debris fractions containing major portions of cell membranes were resolubilized in 8 m urea, 0.5 m DTT by boiling at 95 °C for 20 min. Resolubilized membrane fractions were also subjected to SDS-PAGE. As a control, 0.1 μg of purified recombinant FchR was additionally loaded on the gels, which were developed for 1.2 h at 150 mV, and then proteins were blotted onto a hyperbond PVDF membrane for 1 h at 200 mV. After blocking with 2.5% (w/v) BSA in TBS, polyclonal FchR-specific antibody (Pineda, Berlin, Germany) was applied for 12 h, and after washing secondary antibody (goat anti-rabbit IgG coupled with alkaline phosphatase; Bio-Rad) was applied for 2 h. Blots were developed using 5-bromo-4-chloro-3-indolyl phosphate and nitro blue tetrazolium at pH 9.5 for signal detection.

Chemical Iron and Sulfide Determination

For iron determination, three samples (5, 10, and 20 μl) of 80 μm purified FchR, two buffer controls, and five samples (1, 2, 5, 10, and 20 nmol) of iron standard (NH₄)₂Fe(SO₄)₂·6 H₂O were diluted to 100 μl with H₂O. Subsequently, 100 μl of 1% (w/v) hydrochloric acid were added, and samples were mixed gently. Samples were incubated at 80 °C for 10 min, and 500 μl of 7.5% (w/v) ammonium acetate, 100 μl of 4% (w/v) ascorbic acid, 100 μl of 2.5% (w/v) SDS, and 100 μl of iron chelator (Ferene) were added sequentially. Samples were centrifuged for 5 min at 15,000 × g, and absorbance was measured at 590 nm against water. For determination of acid-labile sulfide, three samples (10, 20, and 40 μl) of 120 μm purified FchR, two buffer controls, and six samples (2, 10, 20, 30, 40, and 50 nmol) of Li₂S sulfide standard were diluted to 200 μl with H₂O. Then 600 μl of 1% (w/v) zinc acetate and 50 μl of 7% (w/v) NaOH were added, and samples were mixed shortly and incubated for 15 min at 22 °C. After short centrifugation, 150 μl of N,N′-dimethyl-p-phenylenediamine, 0.1% (w/v) in 5 m HCl, and 150 μl of 10 mm FeCl₃ in 1 m HCl were added quickly to start methylene blue formation. After short vortexing, samples were centrifuged for 5 min at 15,000 × g, and absorbance was measured at 670 nm against water.

EPR Analysis

Purified FchR was reduced anaerobically by incubation with 5 mm dithionite for 5 min to achieve quantitative cluster reduction. 50 μm of reduced protein were then measured at 40 K and 9.4681 GHz using a Bruker EPR spectrometer. The cluster concentration was quantified by comparing the numerically double-integrated 40 K spectra with those of a 1 mm Cu(II)-EDTA standard without saturation effects.

EPR Redox Titration

20 μm holo-FchR in 10 mm Tris-HCl, pH 7.5, 50 mm NaCl were anaerobically incubated and mixed with redox mediators methyl viologen, benzyl viologen, neutral red, safranine T, phenosafranine, and anthraquinone-2-sulfonate (40 μm each). Actual redox potentials were determined continuously by using a platinum electrode and a Ag/AgCl reference electrode under constant mixing of the protein/indicator solution as described previously (29). Increasing amounts of dithionite were added stepwise, and 13 samples were taken between −215 mV/NHE (corresponding to the start potential) and −520 mV/NHE (corresponding to complete reduction). Immediately, frozen samples were measured at 77 K, and both amplitudes and slopes of the g_y = 1.956 feature were plotted after corrections for sample volumes and concentrations and analyzed by fitting to the Nernst equation.

HPLC and Mass Spectrometric Analysis

Culture supernatant extracts were analyzed by using a C-18 column (Macherey-Nagel) with a water, 0.05% formic acid (A) and acetonitrile, 0.045% formic acid (B) gradient from 5 to 95% B in 30 min with a column temperature of 45 °C. Electrospray ionization-mass spectra of the eluting compounds were recorded in positive ion mode within the mass range of 100–1000 m/z. For a complete desalting of proteins by HPLC using an Agilent 1100 system, samples were applied to a monolithic 50:1 ProSwift RP-4H column (Dionex). Desalted proteins were eluted by the following gradient of A and B at a column temperature of 40 °C and a flow rate of 0.2 ml/min: isocratic elution with 5% A for 2 min, followed by a linear gradient to 95% B within 8 min, and holding 95% B for additional 4 min. Online mass spectrometric analysis was done with a QStar Pulsar i mass spectrometer (ABSciex, Darmstadt, Germany) equipped with an electrospray ionization source. Parameters were as follows: DP1 75, FP 265, DP2 15, CAD 2, GS1 65, and CUR 35. The voltage applied was 5500 V. Positive ions within the mass range of 500–2000 m/z were detected. For better performance, the “Enhance All” mode was activated.

CD Measurements

CD measurements were carried out at a temperature of 20 °C using a J-810 spectropolarimeter (Jasco) and 0.5-cm path length cuvettes. Spectra were recorded with a measuring range from 650 to 265 nm, a bandwidth of 1 nm, a data pitch of 0.2 nm, a response of 1 s, and with standard sensitivity. The scanning speed was 100 nm min⁻¹, and the presented data are an accumulation of 10 scans. The protein concentration of holo-FchR was 20 μm in 20 mm Tris-HCl, pH 8.8 (the presented curves are corrected for the signal of the buffer). First, a spectrum of reduced holo-FchR was recorded, and then Ga(III)-dicitrate was added to a final concentration of 200 μm, and the next measurement was taken, followed by addition of Fe(III)-dicitrate to yield a final concentration of 100 μm and conductance of the next measurement. Incubation times between each addition and measurement were 15 min. The addition of the compounds was performed under anaerobic conditions.

Fluorescence Titration

Purified recombinant protein solution in 150 mm NaCl, 100 mm Tris-HCl, pH 8.0, was adjusted to the desired concentration, and 2 ml were applied into a 1 × 1-cm² quartz cuvette. The cuvette was placed into an FP-6500 spectrofluorimeter (Jasco) and thermostated at 22 °C. Concentrated ferric siderophore stock solutions were freshly prepared and added stepwise to the protein solution. Tyrosine/tryptophan fluorescence was measured after excitation at 280 nm (slit width 5 nm) at 340 nm (slit width 5 nm). Each measurement was averaged three times. Data analysis and K_D calculation were done as described previously (30).

Enzyme Kinetics and Inhibition Studies

All kinetic studies were performed under anaerobic atmosphere (5% H₂ in N₂) at 25 °C. For initial determination of enzyme substrate spectrum, 50 μm FchR was reduced with 5 mm dithionite for 5 min, then purified anaerobically via size exclusion chromatography using an Econo-Pac 10DG column (Bio-Rad), and incubated with 150 μm ferric siderophore substrates (Fe(III)-dicitrate and ferrioxamine E (Sigma); Fe(III)-aerobactin, Fe(III)-schizokinen, ferrichrome (EMC microcollections, Tübingen, Germany), Fe(III)-enterobactin (31)) for 10 min in 100 mm NaCl, 50 mm Tris-HCl with varying pH values at 7.0, 7.5, 7.8, 8.0, 8.3, 8.5, 9.0, and 9.5 (although pH 9.5 is at the limit of the optimal buffer range, the same buffer was used to ensure comparable experimental conditions). Fe/S cluster spectra were measured before and after incubation to monitor cluster reoxidation. To test efficiency of FchR reduction by a physiological electron donor, measurements were repeated by using 2 mm NADH, NADPH, FADH₂ (all obtained from Sigma) and ferredoxin (Fd; source, Spinacia oleracea; Sigma), which was reduced with dithionite and purified anaerobically. Detailed kinetics were then performed in 50 mm Tris-HCl, pH 8.0, 100 mm NaCl at 25 °C by using 1 μm holo-FchR and 10 μm Fd together with a regenerative system consisting of NADPH:Fd oxidoreductase (source, S. oleracea; Sigma), which bears a low potential flavin with E_½ = −442 mV, pH 8.0 (32), and reversibly transfers electrons to Fd, possessing a low potential Fe₂S₂ cluster with E_½ = −420 mV (33), through formation of a 1:1 stoichiometric complex (32). Furthermore, glucose-6-phosphate dehydrogenase (source, S. cerevisiae; Sigma) was used to ensure low steady-state concentrations of NADP⁺ potentially acting as a competitive inhibitor of NADPH:Fd oxidoreductase (34). Nonlimiting rate determination was done with starting concentrations of 2 mm glucose 6-phosphate and 2 mm NADPH, 0.5 units of regenerative enzymes, and varying concentrations of potassium ferricyanide (E′₀, pH 7.0, = 436 mV). Conditions found to be sufficient for nonlimiting electron transfer were then applied to FchR-dependent kinetics with ferric siderophores over a substrate concentration range from 1 to 1000 μm. After 10 min, preincubation of regenerative enzymes with their substrates (glucose 6-phosphate, NADPH, Fd) and FchR, ferric siderophores were added, and starting velocities of iron release were measured by 3 min of incubation (initially determined to be within the time-dependent linear catalytic range). Fe(II) release was detected by addition of 0.15% (w/v) ferene and immediate measurement of absorbance at 590 nm using an Ultrospec 3000 spectrophotometer (GE Healthcare) or, alternatively, a NanoDrop ND-1000 (peQLab). Fe(II) concentrations were calculated by using a Fe(II)-ferene calibration curve measured at the same wavelength. Further kinetics for Fe(III)-dicitrate, Fe(III)-aerobactin, and ferrichrome were performed in presence of 100 μm apo-SufU (source, B. subtilis (35)), which was de-metallized before by three dialysis steps in 100 mm EDTA-containing reaction buffer and frequent thermal shifts to 45 °C. Background reduction of ferric siderophore substrates (at maximum 4% relative to FchR-dependent reduction) was determined for each kinetic in absence of FchR and subtracted from iron release in the presence of FchR. Each kinetic was performed three times independently, and averaged data were plotted with obtained standard deviations. Kinetic parameters were determined upon Michaelis-Menten analysis by using Microcal Origin 5.0 software. Inhibition assays were performed by using Fe(III)-dicitrate as a substrate and Ga(III)-dicitrate and Ga(III)-desferrioxamine E as potential inhibitors. To determine the K_i values, Fe(III)-dicitrate concentrations were set around the K_m value and additionally at saturating concentrations. Concentrations of the inhibitors were varied from 0 to 70 μm in case of Ga(III)-dicitrate and 0 to 12.5 μm in case of Ga(III)-desferrioxamine E. Assays were performed according to the established standard kinetic conditions. The inhibition constants were calculated by using the Dixon plot method (36).

Mutant Construction

A PCR hybrid construct was generated by fusing the Cm^R cassette of vector pX (37) with homologous flanking regions of gene BH1040 (38). The 3′-ends of the upstream and downstream flanks contained complementary 3′-ends to the resistance cassette fused in a second PCR. PCRs were performed with Platinum®Pfx DNA polymerase (Invitrogen) and the Expand long template PCR system (Roche Applied Science). Transformation was done with B. halodurans protoplast cells following described methods (39, 40). Shortly thereafter, protoplasts were generated in Protoblast(PB)-buffer containing 20% (w/v) sucrose, 10 mm MgCl₂, and 20 mm Tris-HCl, pH 9.2, supplemented with 5% (w/v) lysozyme at 37 °C. They were harvested by centrifugation at 3000 × g for 10 min and resuspended in PB-buffer. DNA was added together with 25% (w/v) PEG 8000, and after 10 min of incubation, 2× Penassay broth containing 20% (w/v) sucrose was added with equal volume, and protoplasts were recovered by centrifugation. Treated protoblasts were resuspended in Penassay broth-mixed PB-buffer and gently shaken for 30 min at 30 °C. Transformants were selected on modified DM3 regeneration plates with chloramphenicol at 20 μg/ml.

Growth Assays and Siderophore Tests in Defined Medium

For defined growth studies with B. halodurans WT and ΔfchR, modified iron-limited Belitsky minimal medium (41) at pH 9.5 was used. Strains were inoculated to a starting A₆₀₀ of 0.05, and growth was monitored continuously until stationary phase was reached. Endogenous siderophore production was tested by Arnow (42) and modified Csaky test (43). Extraction of culture supernatants for analysis of siderophore species was done as described previously (44). For siderophore supplementation assays, microtiter scale cultures were used that were supplemented with 100 μm of FeCl₃, ferric siderophores, or Ga(III)-loaded siderophores. Cultured were incubated at 300 rpm at 37 °C, and growth was monitored by using a microtiter plate reader. Three independent cultures for each condition were set up, and growth data were plotted with their corresponding standard deviations.

ICP-MS Analysis

Cultures were inoculated with overnight cultures to a starting A₆₀₀ of 0.05 and were harvested after reaching stationary growth phase. Cells were centrifuged at 18,000 × g for 5 min; pellets were washed three times with buffer containing 10 mm Tris-HCl, 1 mm EDTA, pH 7.5, and finally with milliQ/water to remove extracellular traces of salt. Cells were dried for 20 h at 100 °C and treated with suprapure nitric acid for quantitative breaking, and intracellular metal contents were analyzed by ICP-MS using an ICP-MS Agilent 7500ce. Three biological replicates for each condition were analyzed, and averaged data of the measurements were given with corresponding standard deviations. Analysis of metal contents in protein solutions (20 mm Tris-HCl, pH 7.5) was done after 1:20 dilution appropriate to obtain the required sample volumes and identical sample viscosities. Yttrium was added to all samples as an internal standard.

RESULTS

Identification of Iron-associated Reductase Genes in Firmicutes

Several species within the Firmicutes group possess putative non-flavin reductase genes with conserved C-terminal cysteines that are closely associated with ATP-binding cassette-type iron uptake systems (Fig. 1, A and B). A phylogenetic analysis shows that these Firmicute reductases form a rather heterogeneous but distinct cluster clearly distinguished from the homogeneous cluster of enterobacterial FhuF-like reductases that show a high degree of conservation among each other (Fig. 1C). Transmembrane-spanning segments predicted within the FhuF sequences are not or with high uncertainty predicted within the Firmicute reductase sequences, suggesting topological differences between these reductases of different phyla. Within the Bacillus genus, the conserved C-terminal CCX₄CX₆CX₂C motif within the reductase sequences partially differs from the CCX₁₀CX₂C FhuF motif, and the identities of the full-length amino acid sequences compared with E. coli FhuF are between 4 and 14%, indicating that the homology of the corresponding genes is uncertain. This suggests that the predicted Firmicute reductases represent a novel type of putative FSRs with C-terminal conserved cysteine motif potentially showing novel features compared with FhuF.

FIGURE 1.

A, global ClustalW alignment (identical and similar amino acids shown black and gray shaded boxes, respectively) of putative ferric siderophore reductases with C-terminally conserved cysteine motifs (highlighted in orange) from five Firmicute species (Gsp, Geobacillus sp. WCH70; Afl, Anoxybacillus flavithermus WK1; Bli, Bacillus licheniformis ATCC14580; Bme, B. megaterium DSM319; Bha, B. halodurans C-125) and five Enterobacteriaceae (Cro, Citrobacter rodentium ICC168; Sen, Salmonella enterica subsp. enterica SPB7; Sso, Shigella sonnei Ss046; Eco, E. coli K-12; Kpn, Klebsiella pneumoniae 342). Predicted transmembrane regions in the FhuF-type reductases are indicated by red-coded box. B, association of ferric siderophore ATP-binding cassette-type uptake genes and putative reductases in Firmicute species. Blue, substrate binding genes; yellow, permease genes; green, ATPase genes; red, reductase genes. Nomenclature for B. licheniformis and B. megaterium genes was taken from homologs of the B. subtilis ferric hydroxamate transporter yfiY-yfiZyfhA-yusV (51). C, phylogenetic clustering of reductase sequences aligned in A. Theoretical evolutionary rates of nucleotide substitutions resulting in sequence diversity by assuming one common ancestor are indicated. D, upper panel, predicted gene cluster for schizokinen biosynthesis (black arrows) and export (gray arrow) from B. megaterium QM B1551 and B. halodurans C-125 with given GenBank^TM locus tags and percentages of amino acid sequence identities. Predicted gene functions are as follows: BMQ_4069/BH2624 (S. meliloti RhbA homologs), 2,4-diaminobutyrate 4-transaminase; BMQ_4068/BH2623 (RhbB homologs), l-2,4-diaminobutyrate decarboxylase; BMQ_4067/BH2622 (RhbC homologs); type A siderophore synthetase; BMQ_4066/BH2621 (RhbD homologs), acyl-CoA transferase; BMQ_4065/BH2620 (RhbE homologs), type B siderophore synthetase; BMQ_4064/BH2619, major facilitator superfamily-type transporter; BMQ_4063/BH2618 (RhbF homologs), type C siderophore synthetase. Lower panel, electrospray ionization-MS positive ion mode spectrum of extracted schizokinen from B. halodurans iron-deprived culture medium. Mass peaks are as follows: 403.1 = [M − H₂O + H]⁺; 421.4 = [M + H]⁺; 424.9 = [M − H₂O + Na]⁺; 443.2 = [M + Na]⁺; 456.2 = [M − H₂O + Fe − 2H]⁺; 459.1 = [M + K]⁺; 473.9 = [M + Fe − 2H]⁺.

Among known Bacillus species, B. halodurans is one exception because it lacks genes to synthesize or utilize catecholate siderophores. Although rather closely related to the nonalkaliphilic model bacterium B. subtilis (45), the complete bacillibactin pathway is absent in the genome, including biosynthesis genes dhbACEBF, ferric bacillibactin uptake genes feuABC, and trilactone hydrolase besA. However, uptake systems for elemental iron, iron dicitrate, and iron hydroxamates are present and similar to the corresponding systems in B. subtilis. To test siderophore production in B. halodurans, iron-limited minimal medium cultures were grown until stationary phase, and secretion of catecholate and/or hydroxamate compounds was tested by using the Arnow and Csaky tests, respectively. Only the Csaky test specific for hydroxamates gave a positive response (data not shown), revealing the ability of B. halodurans to produce hydroxamic compounds during iron limitation, which is in agreement with a recent study (46). HPLC-MS analysis of extracted iron-deprived cultures revealed the presence of a mass pattern according to schizokinen (Fig. 1D) (47). This citrate-hydroxamate siderophore, which is structurally closely related to aerobactin produced by Enterobacteriaceae, was so far only isolated from Bacillus megaterium (48, 49). However, the present genome analysis revealed in both B. megaterium and B. halodurans the same gene cluster with high homology that includes all genes found to be involved in rhizobactin 1021 biosynthesis in Sinorhizobium meliloti except rhbG, which is required for asymmetric acylation of the otherwise schizokinen-like scaffold (Fig. 1D) (50). Thus, B. halodurans was identified as a further species producing schizokinen during iron limitation, further supporting the requirement for a reductive iron release mechanism in this bacterium.

Sequence analysis of the reductase-associated transport system revealed homology to B. subtilis Fe(III)-siderophore ATP-binding cassette transporters. Highest identities are shared with the Fe(III)-dicitrate YfmCDEF and the ferric citrate-hydroxamate YfiYZ-YfhA-YusV transporter (51) by B. halodurans ORFs BH1037 (29 and 26% amino acids identity to substrate-binding proteins YfmC and YfiY, respectively), BH1038 (37% amino acids identities for both YfmD and YfmE and 34 and 32% amino acids identities for YfiZ and YfhA to the N- and C-terminal sequence parts of the BH1038 transmembrane fusion protein, respectively), and BH1039 (56 and 74% identities to ATPase subunits YfmF and YusV, respectively). In contrast to B. subtilis, the B. halodurans gene BH1039 is associated with an additional gene, BH1040, encoding the putative FSR. The overlapping start and stop codons of genes BH1039 and BH1040 are indicative for their translational coupling, which has also been observed for post-transcriptional regulation of fur expression (52); however, it is not a common regulatory feature in the context of bacterial iron metabolism. Thus, BH1040 is predicted to encode a novel type of FSRs in Firmicutes and hence would represent the first example of a non-flavin FSR in Gram-positive bacteria.

Production and Initial Characterization of the BH1040 Gene Product

To address the function of BH1040, the gene was recombinantly expressed as a Strep-tag II fusion, and the protein was purified via Streptactin chromatography. The reddish-brown color of concentrated elution fractions pointed to the binding of a chromogenic cofactor. The UV-visible spectrum of the aerobically purified protein showed in addition to 280 nm significant absorption features at 330 and 450 nm as well as slight additional peaks at 560 and 670 nm, which were strongly weakened upon dithionite reduction, indicating the binding of an Fe/S cluster (Fig. 2A). EPR spectroscopy revealed that the cluster was EPR-silent in its oxidized state. Upon reduction with 5 mm dithionite for 5 min, an EPR signal indicative for the presence of a reduced [2Fe-2S]⁺ cluster was found with g values g_z = 2.001, g_y = 1.956, and g_x = 1.866 (Fig. 3A). These values are similar to those reported for FhuF (25), but the total FchR EPR spectrum is broader, and thus g_z and g_x are ranged above and below the corresponding FhuF g values, respectively. By using a 1 m Cu(II)-EDTA standard, the cluster-binding holoprotein fraction was estimated with 25 μm in a total protein concentration of 55 μm. Thus, about half of the fraction of the recombinantly purified protein was found to be loaded with the Fe/S cofactor. These findings were supported by mass spectrometric analysis revealing the presence of two fractional species at nearly equal ratios (Fig. 2B). One species showed the mass expected for the recombinant apoprotein, although the other species showed a mass shift of plus 174 corresponding to the presence of a [2Fe-2S] cluster. This indicates that one BH1040 monomer is able to bind one complete Fe/S cofactor. Chemical determination of releasable iron and sulfide revealed equal molar ratios of both compounds, which were further found to be in molar stoichiometries of 1:1 with total purified protein. Given a holoprotein fraction of nearly 50%, this indicated the binding of two labile iron and sulfur species per holoprotein monomer. The same molar ratio of iron to holoprotein content was obtained by ICP-MS analysis of an BH1040 dilution series over a range of 10⁰ to 10⁵ ppb. To further determine a possible oligomerization state of the protein, analytical gel filtration was performed, revealing that the protein dominantly exists in the monomeric state (Fig. 2C). The combination of apo-BH1040 and holo-BH1040 did not result in different retention signals, indicating that cluster binding does not significantly change the hydrodynamic properties of BH1040. To further analyze the putative iron-dependent regulation and the cellular topology, polyclonal antibodies were raised and used for Western blot analysis of total cellular protein of cytosolic and membrane-containing fractions from cultures grown under iron-sufficient and iron-deprived conditions (Fig. 2D). As a result, the protein was detected in cytosolic fractions of cultures treated with each 5 or 50 μm of both 2,2′-bipyridyl and 1,10-phenanthroline demonstrating its iron-dependent induction. No signal was detected in membrane fractions even in presence of each 75 μm 2,2′-bipyridyl and 1,10-phenanthroline, suggesting that BH1040 is a cytosolic protein and not associated with the cytosolic membrane.

FIGURE 2.

Cofactor characterization, iron-dependent regulation, and cellular localization of BH1040 (FchR). A, UV-visible analysis of FchR (35 μm) after aerobic purification and dithionite reduction; inset, SDS-PAGE analysis of purified protein. B, mass spectrometric analysis of aerobically purified FchR with recombinant Strep-tag II showing the mass of apo-FchR-Strep-tag II (32,585 atomic mass units) and holo-FchR-Strep-tag II (32,759 atomic mass units) carrying a [2Fe-2S] cofactor. C, analytical gel filtration using a Zorbax GF-250 column with 100 μg of FchR and calibration proteins; F, ferritin; A, aldolase; C, conalbumin; O, ovalbumin; CA, carbonic anhydrase; AP, aprotinin. D, Western analysis with FchR polyclonal antibodies. Lane 1, marker; lane 2, recombinantly purified FchR; lane 3, B. halodurans cytosolic protein extract; lane 4, B. halodurans ΔfchR cytosolic protein extract (50 μm BP/50 mm Phen); lane 5, B. halodurans cytosolic protein extract (5 μm BP/5 μm Phen); lane 6, B. halodurans cytosolic protein extract (50 μm BP/50 μm Phen); lane 7, B. halodurans membrane fraction; lane 8, B. halodurans membrane fraction (50 μm BP/50 μm Phen).

FIGURE 3.

EPR analysis of BH1040 (FchR) Fe/S cofactor and EPR redox titration. A, EPR spectrum of dithionite-reduced FchR measured at 40 K with 9.4681 GHz microwave frequency, 0.2 milliwatt microwave power, 100 kHz modulation frequency, and 1.25 mT modulation amplitude. Found g values are as follows: g_z = 2.001; g_y = 1.956; g_x = 1.866. B, normalized Nernst plot of EPR amplitude change (based on the 1.956 EPR feature) obtained from redox titration with dithionite showing a midpoint potential of −348.4 mV (versus NHE). Spectra were obtained at 77 K with 9.4681 GHz microwave frequency, 5.0 milliwatt microwave power, 100 kHz modulation frequency, and 1.25 mT modulation amplitude.

Determination of the [2Fe-2S] Cluster Redox Potential

To determine the Fe/S cluster midpoint redox potential as an estimate of the redox capacity of holo-BH1040, the protein was subjected to anaerobic EPR titration using dithionite as the reducing agent. First, it was tested if the Fe/S cluster was stable against elongated incubation with dithionite. EPR analysis revealed complete stability of the [2Fe-2S]⁺ signal over 45 min of continuous dithionite reduction at pH 7.5, and thus, redox titration was performed at same pH within this time frame. By following the amplitude changes of the g_y = 1.956 feature, a midpoint potential of −348.4 mV (versus NHE) was determined by Nernst fitting (Fig. 3B). EPR slopes were further evaluated, resulting in essentially the same potential. Thus, FchR was revealed to possess a midpoint potential lower than free flavin or nicotinamide cofactors and even lower than determined for the E. coli FhuF reductase (26).

Characterization of the Substrate Spectra of BH1040 and BH1037

To issue the capacity of BH1040 to release iron from ferric siderophores, the protein was reduced with 5 mm dithionite under anaerobic conditions. Dithionite was removed by size exclusion chromatography, and reduction of the [2Fe-2S] cluster was proofed by UV-visible spectroscopy. The reduced holoprotein was then incubated with several ferric siderophores for 10 min at pH 8.0, and both the UV-visible spectrum of the Fe/S cofactor and the amount of released Fe(II) were analyzed. In case of Fe(III)-dicitrate and several ferric hydroxamates, the UV-visible spectrum of the cluster was found to shift into the oxidized state, but not in case of ferric triscatecholates (Fig. 4A). Similarly, the release of Fe(II) detected by ferene absorbance at 590 nm was most strong with Fe(III)-dicitrate (100%), significantly present with Fe(III)-aerobactin (47%) and ferrichrome (16%), but less than 0.1% with Fe(III)-enterobactin. Thus, because of its capability of releasing iron from Fe(III)-dicitrate, ferric citrate-hydroxamates, and ferric hydroxamates, the protein was named consequently FchR (ferric citrate and hydroxamate reductase). The highest enzyme activity was found between a pH range of 7.8 and 8.5, although lower activities were found at pH 7.0 and 9.5 with 87 and 73% of release from Fe(III)-aerobactin compared with pH 8.0 at the same experimental conditions. Thus, the enzyme showed an optimal activity at modest alkaline conditions in vitro, which corresponds to the measured internal pH of alkaline bacteria, including B. halodurans that is maintained at around 8.0 in the presence of an external pH from 8.0 to 11.0 (53). To further address the specificity of the associated substrate-binding protein BH1037, the protein was recombinantly produced as Strep-tag II fusion, and binding of several ferric siderophores was tested by tyrosine/tryptophan fluorescence quenching (Fig. 4B). Although no significant binding was observed with Fe(III)-dicitrate, ferrichrome, and ferric triscatecholates, titration with Fe(III)-schizokinen and Fe(III)-aerobactin led to strong quenchings resulting in similar dissociation constants of 3.9 ± 1.5 and 8.0 ± 4.4 nm, respectively. Thus, BH1037 showed strong substrate specificity for the structurally related ferric citrate-hydroxamates Fe(III)-schizokinen and Fe(III)-aerobactin, suggesting that the transporter associated with FchR delivers this class of preferred FchR substrates into the cytosol.

FIGURE 4.

A, qualitative estimation of substrate spectrum of FchR by incubation of 50 μm reduced and purified enzyme with 150 μm of potential ferric siderophore substrates for 10 min at pH 8.0. UV-visible spectra were measured before and after incubation to monitor Fe/S cluster reoxidation. B, fluorescence titration of BH1037 applied at a concentration of 3 μm was performed with Fe(III)-dicitrate, ferrichrome, Fe(III)-schizokinen, and Fe(III)-aerobactin. Quenching curves obtained with Fe(III)-schizokinen and Fe(III)-aerobactin upon excitation of protein tyrosine/tryptophan fluorescence at 280 nm, and emission of fluorescence at 340 nm were fitted according to the law of mass action by using a 1:1 stoichiometric binding model. C, scheme of electron donor/recycling system coupled with FchR-dependent activity for kinetic analysis of iron release (upper panel). Applied starting concentrations were 2 mm glucose 6-phosphate, 2 mm NADPH, and 10 milliunits of regenerative enzymes ferredoxin:NADP⁺ reductase and glucose-6-phosphate dehydrogenase. After 10 min of equilibration, transfer rates for nonlimiting electron shuttling in the presence of varying concentrations of ferricyanide as terminal acceptor were determined at pH 8.0 (lower panel). Data of three independent determinations were averaged, plotted with standard deviations, and analyzed by Michaelis-Menten kinetic. D, iron release kinetics with different ferric siderophore substrates and pre-equilibrated 1 μm FchR, 10 μm Fd, and electron donor/recycling system as indicated in C. Release activity was monitored by determination of Fe(II)-ferene absorbance at 590 nm. Amounts of released Fe(II) per time were calculated using a Fe(II)-ferene spectral calibration curve, and background activity was subtracted for each measurement by performing control reactions without FchR. Data of three independent determinations for each substrate were averaged, plotted with corresponding standard deviations, and fitted according to the Michaelis-Menten model.

Kinetic Analysis of Iron Release Using a Ferredoxin-coupled Electron Donor/Recycling System

To reduce the cofactors of FSRs, coupled processes of electron transfer are necessary. It is known that reduced flavin cofactors such as FMNH₂ and FADH₂ are released from corresponding NAD(P)H:FMN reductases or NAD(P)H:FAD reductases and nonspecifically reduce siderophores, especially under conditions of low pH (54). However, it was not shown yet how an FSR that interacts directly with the ferric siderophore substrates is reduced by upstream electron transfer processes. Thus, before recording detailed kinetics of iron release, we searched for a native electron donor system that would ensure the nonlimited catalytic function of FchR in vitro. Initial tests with several reduced cellular electron carriers, including FMNH₂, FADH₂, NAD(P)H, ascorbate-H₂, and Fd, revealed that Fd_red led to efficient reduction of the FchR [2Fe-2S] cluster revealed by UV-visible spectral analysis. Tests with several ferric siderophore substrates showed that fast iron release occurred when incubating oxidized FchR with reduced Fd, whereas reduced Fd alone had a background activity of maximal 4% relative to FchR during a reaction time of 1 min. Because the genome of B. halodurans contains at least three ferredoxin homologs (BH0209, BH1605, and BH2357) and furthermore a putative ferredoxin:NADP⁺ reductase (BH3408, a homolog to B. subtilis yumC), a three-component electron donor system was established consisting of NADPH as primary source, Fd as FchR reducer, and a ferredoxin:NADP⁺ reductase as a recycling enzyme (Fig. 4C). Regeneration of NADPH for high steady-state activities was ensured by coupling the ferredoxin:NADP⁺ reductase with glucose-6-phosphate dehydrogenase. Electron transfer rates of this system were tested with ferricyanide as a high potential substrate and were found to have a k_cat of 398 ± 2 s⁻¹ (Fig. 4C). The high rate electron transfer was then applied to perform FchR-dependent kinetics with various potential ferric siderophore substrates. All kinetics followed a Michaelis-Menten-type behavior (Fig. 4D), and kinetic parameters could be determined for Fe(III)-dicitrate and Fe(III)-(citrate)-hydroxamate conversion (Table 1). The triscatecholate Fe(III)-enterobactin proved to be a poor substrate, and the conversion rate was near background activity without observed saturation, and thus the k_cat value was estimated to be below 0.1 min⁻¹, although the K_m value was not specified further. All further kinetics were saturated, although with very different k_cat and K_m values changing drastically with increasing redox potential of the free ferric chelate substrates. Only for Fe(III)-schizokinen and Fe(III)-aerobactin were similar kinetic parameters obtained, accounting for their structural and hence redox potential similarity. Significant catalytic efficiencies for low potential ferric hydroxamates such as ferrichrome and ferrioxamine E were only observed because their K_m values strongly dropped, indicating stronger substrate affinity despite lower turnover rates. Because k_cat values for those substrates were quite low under the tested conditions, and the redox equilibrium favored the state of Fe(III) because of the presence of released apo-siderophores, we tested if the essential Gram-positive scaffold protein SufU (BH3468 homolog) from the related species B. subtilis (35, 55), which tightly binds Zn(II) (∼0.75 mol/mol protein) when recombinantly produced, would enhance the release of Fe(II) by shifting the iron redox equilibrium to the ferrous side. Indeed, de-metallized apo-SufU that contained only 0.05 mol of Zn(II)/mol of protein had strong effects on both K_m and k_cat values for assayed ferric substrates. Catalytic efficiencies were found to be enhanced 2.2-, 5.8-, 5.4-, and 1.3-fold in presence of 100 μm apo-SufU in the case of Fe(III)-dicitrate, Fe(III)-schizokinen, Fe(III)-aerobactin, and ferrichrome, respectively (Table 1). These data indicate that efficient product scavenging can be a further strategy in vivo to enhance reduction efficiency of the available iron sources.

TABLE 1.

Kinetic data of FchR for various ferric siderophore substrates determined anaerobically in 50 mm Tris-HCl, pH 8.0, 100 mm NaCl at 25 °C

ND means not determined.

	Fe(III)-dicitrate	Fe(III)-schizokinen	Fe(III)-aerobactin	Ferrichrome	Ferrioxamine E	Fe(III)-enterobactin
Km (μm)	110 ± 8	93 ± 4	96 ± 24	68 ± 10	26 ± 2	ND
Km (μm) in presence of 100 μm apo-SufU	94 ± 6	49 ± 5	55 ± 5	52 ± 6	ND	ND
kcat (min−1)	114 ± 3	32 ± 0.4	27 ± 2	11 ± 0.5	2 ± 0.1	<0.1
kcat (min−1) in presence of 100 μm apo-SufU	214 ± 7	98 ± 7	84 ± 5	26 ± 3	ND	ND
kcat/Km (mm−1 s−1)	17.2	5.8	4.7	2.7	1.5	ND
kcat/Km (mm−1 s−1) in presence of 100 μm apo-SufU	37.8	33.5	25.1	8.3	ND	ND
Specific activity (units/mg)	3.5	1.0	0.8	0.3	0.1	ND
Specific activity (units/mg) in presence of 100 μm apo-SufU	6.6	3.0	2.6	0.8	ND	ND
E′0 (pH 7.0) (mV) of free ligand	∼ 0a	ND	-336b	-400c	-481d	-750e

^a Data are from Ref. 6.

^b Data are from Ref. 77.

^c Data are from Ref. 78.

^d Data are from Ref. 79.

^e Data are from Ref. 17.

Enzyme Inhibition with Ga(III)-Citrate and Ga(III)-Ferrioxamine E

Because Fe(III) can be efficiently replaced in siderophores with redox inert Ga(III) because of their similar coordination properties (56), we tested if such siderophore mimics would inhibit FchR-mediated ferric siderophore conversion. The standardized kinetic conditions were applied to inhibition tests using Fe(III)-dicitrate as substrate and Ga(III)-dicitrate and Ga(III)-ferrioxamine E as potential inhibitors. Initial studies revealed minor inhibitory effects for Ga(III)-dicitrate, although effects for Ga(III)-ferrioxamine E were significantly stronger. Because inhibitory effects were strictly dependent on the present substrate concentrations, competitive inhibition of Ga(III)- and Fe(III)-charged siderophores at the FchR active site was predicted. The competitive inhibition mode was verified during detailed inhibition studies with variations of inhibitor and substrate concentrations in the initially observed inhibition range. To determine inhibition constants for the two inhibitors, data of the inhibition series were analyzed by Dixon plot method resulting in a K_{i(Ga(III)-dicitrate)} of 86 μm and a K_{i(Ga(III)-ferrioxamine E)} of 14.5 μm (Fig. 5).

FIGURE 5.

Inhibition studies with Ga(III)-charged redox-inert substrate mimics. By applying the standard kinetic conditions, inhibition assays were performed with Fe(III)-dicitrate as FchR substrate and Ga(III)-dicitrate (A) and Ga(III)-desferrioxamine E (B) as potential inhibitors. To determine inhibition constants, different Fe(III)-dicitrate concentrations were chosen (50, 100, 200, and 400 μm), and concentrations of the inhibitors were varied from 0 to 70 μm in case of Ga(III)-dicitrate and 0 to 12.5 μm in case of Ga(III)-desferrioxamine E. Three independent measurements for each concentration were performed; data were averaged and plotted with their standard deviations according to the Dixon plot method. K_i values were determined from corresponding curve intersections of the plotted diagrams.

Direct Binding Studies with Ferric Siderophores and FchR

Because the kinetic analysis revealed a correlation between catalytic efficiency and siderophore redox potential, we attempted to analyze if FchR is able to specifically recognize different ferric siderophores characterized by octahedral oxygen-mediated iron coordination, and next if binding of the metal-charged siderophore would alter the geometry of the catalytically active Fe/S cofactor in a way that inner sphere electron transfer could occur. To directly measure protein-ligand interactions between FchR and different substrates, catalytically inactive FchR bearing the oxidized Fe/S cofactor was used to determine binding constants by fluorescence titration. Iron-charged and noncharged forms of citrate, aerobactin, desferrichrome, and desferrioxamine E were used as ligands in titration experiments. Although noncharged forms did not display binding within the micromolar range, the tested ferric iron complexes showed K_D values between ∼10 and 100 μm with 18 ± 1 μm for ferrioxamine E, 58 ± 4 μm for ferrichrome, 74 ± 5 μm for Fe(III)-aerobactin, and 104 ± 8 μm for Fe(III)-dicitrate (Fig. 6A). Thus, the determined binding affinities for these substrates were found to be in a similar range within the same order of magnitude. This suggests that a common motif of these complexes is responsible for ligand-protein interaction. Because no binding of nonloaded siderophores was observed, formation of the octahedral iron-oxo motif seems to be essential for the substrate-dependent interactions.

FIGURE 6.

A, fluorescence titrations with oxidized (catalytically inactive) holo-FchR and various ferric siderophore substrates. Concentrations of 50 μm FchR were used for titration with Fe(III)-dicitrate, Fe(III)-aerobactin, and ferrichrome, and 20 μm FchR were used for titration with ferrioxamine E. Quenching curves obtained after tyrosine/tryptophan excitation at 280 nm by fluorescence emission at 340 nm were fitted to the law of mass action by assuming 1:1 stoichiometric binding. B, CD analysis of the [2Fe-2S] cluster spectrum of holo-FchR, either after aerobic purification or after quantitative reduction of the enzyme. Additionally, Ga(III)-dicitrate and Fe(III)-dicitrate were sequentially added to reduced holo-FchR under anaerobic conditions after each measurement to compare spectral changes upon redox-inert and redox-active substrate interaction.

In addition to the determination of enzyme-ligand binding affinities, the possibility of Fe/S cluster rearrangement upon substrate binding was tested, which may occur upon occupation of the inner coordination sphere of the metal-oxo center during electron transfer. For this purpose, reduced FchR with or without additional dithionite was incubated anaerobically with excess Ga(III)-dicitrate (500 μm) and varying amounts (0–200 μm) Fe(III)-dicitrate, which would allow us to trap an interactive transition state between cofactor and substrate. Mixtures were analyzed by EPR and CD spectroscopy. In the case of added Ga(III)-dicitrate, no significant changes of signal pattern compared with the reduced FchR spectrum were observed. Further anaerobic addition of Fe(III)-dicitrate to the mixture of FchR and Ga(III)-dicitrate resulted in stepwise oxidation without changing the EPR-monitored Fe/S cluster signal. Addition of Fe(III)-dicitrate, which exceeded the reduced FchR concentration in the presence of Ga(III)-dicitrate, led over a time period of 15 min to complete oxidation of the cluster, as monitored by CD (Fig. 6B), showing that the catalytic properties of FchR remained fully intact. Thus, because cluster geometry was not found to be affected in the presence of both redox-inert and redox-active metal-oxo centers, we infer no direct interaction between Fe/S cofactor and siderophore substrate.

Growth Analysis of an fchR Deletion Mutant

To test the in vivo effect of FchR deficiency, a B. halodurans fchR deletion mutant was constructed by replacing the gene via homologous recombination by a chloramphenicol resistance cassette upon protoplast transformation. The obtained mutant strain was subjected to growth studies in defined minimal medium. Cell density of ΔfchR under iron-limited conditions after entry into stationary phase was about 25% of wild-type (WT) culture. However, addition of 100 μm Fe(III) restored growth to about 87% compared with WT under iron repletion (Fig. 7A). In contrast, feeding of iron-limited WT and mutant cultures with 100 μm Fe(III)-dicitrate, Fe(III)-aerobactin, ferrichrome, and ferrioxamine E led to a strongly reduced mutant growth recovery that was not higher than about 30% in comparison with equally treated WT cultures. These data suggested that FchR is essential for iron acquisition from both citrate and hydroxamate siderophores. We further tested the effect of redox-inert siderophore mimics Ga(III)-dicitrate, Ga(III)-aerobactin, and Ga(III)-desferrioxamine E on iron-limited WT and ΔfchR cultures. Supplementation of 100 μm of these compounds decreased WT growth to 1.7-, 2.3-, and 1.3-fold, respectively, whereas mutant growth was not significantly affected (Fig. 7B). These data indicate that Ga(III)-loaded siderophores affect the FchR-dependent ferric siderophore reduction and may act as competitive inhibitors of FchR in the bacterial cultures.

FIGURE 7.

Growth analysis of B. halodurans WT (dark gray bars) and ΔfchR (light gray bars) in defined minimal medium supplemented with redox-active or redox-inert FchR substrates. A, cells were grown in iron-limited minimal medium without addition of iron (−Fe(III)) or with addition of either 100 μm FeCl₃ (Fe(III)), Fe(III)-dicitrate (Fe(III)-DC), Fe(III)-aerobactin (Fe(III)-AB), ferrichrome (Fe(III)-FC), or ferrioxamine E (Fe(III)-FO). Absorbances (A₆₀₀) of stationary cultures were monitored from three independent growth experiments, and averaged data were plotted with corresponding standard deviations. B, cells were grown in modified iron-limited Belitsky minimal medium without addition of iron or gallium (−Fe(III)/Ga(III)) or with addition of either 100 μm GaCl₃ (Ga(III)), Ga(III)-dicitrate (Ga(III)-DC), Ga(III)-aerobactin (Ga(III)-AB), or Ga(III)-desferrioxamine E (Ga(III)-DFO). Cultures of three independent growth experiments each were grown to stationary phase, and averaged absorbances (A₆₀₀) were plotted with corresponding standard deviations.

ICP-MS Analysis Reveals That the fchR Mutant Accumulates Iron Intracellularly

Because the fchR mutant showed a strongly reduced growth during iron deprivation, we analyzed the metal content of the intracellular fractions of WT and mutant cells by using ICP-MS (Table 2). Strikingly, the fchR mutant was found to contain a 4.4-fold higher intracellular iron content than the WT. Under iron repletion (100 μm FeCl₃), the relative cytosolic iron level in ΔfchR shifted to 1.2-fold compared with the WT and thus was still higher even under conditions of extracellular iron excess. Although addition of 1 mm citrate to iron-repleted cultures during mid-log growth phase still increased WT growth compared with iron-repleted cultures not supplemented with citrate, growth of mutant cultures treated in the same way was not significantly affected, but intracellular iron content further increased 4.5-fold. To address if intracellular accumulations of iron in the fchR mutant was dependent on the presence of ferric siderophore species, iron-limited cultures of WT and ΔfchR supplemented with 100 μm Fe(III)-dicitrate, Fe(III)-aerobactin, ferrichrome, and ferrioxamine E (see Fig. 7A) were subjected to total cellular iron determination. Although supplementation of ferric siderophores led to an intracellular increase of iron content in the WT in the range of 2.5–7.5-fold, the increase was drastically higher in the mutant ranging from about 4.0- to 13.0-fold compared with nonsupplemented cultures. In contrast, supplementation with Ga(III)-charged siderophores (see Fig. 7B) led to a comparable accumulation of gallium in both mutant and WT, although at a lower level compared with the iron accumulation (Table 2). Dependent on the chelator, intracellular accumulation of iron and gallium strongly varied, thus indicating that especially ferrioxamine E and Ga(III)-desferrioxamine E uptake rates were slower than those observed for Fe(III)-dicitrate or Fe(III)-aerobactin. In conclusion, FchR was found to be an essential component of Fe(III)-dicitrate and ferric (citrate)-hydroxamate utilization in iron-limited cell cultures. Its absence in ΔfchR led to distortion of iron homeostasis associated with strong cytosolic iron accumulation, and its presence in WT made cells susceptible for growth inhibition caused by Ga(III)-charged substrate mimics.

TABLE 2.

ICP-MS-determined intracellular metal abundances given in ppm based on dried cell weight

For analysis, B. halodurans WT and ΔfchR cells were grown in iron-limited minimal medium at pH 9.5 to stationary phase with or without addition of iron or gallium compounds. In line three, 1 mm citrate was added during mid-log growth phase to cultures supplemented with 100 μm FeCl₃.

Culture condition	WT	ΔfchR
−Fe(III)/−Ga(III)	46 ± 4	199 ± 5
100 μm Fe(III)	323 ± 8	388 ± 7
+1 mm citrate	317 ± 5	1729 ± 55
100 μm Fe(III)-dicitrate	331 ± 12	2345 ± 68
100 μm Fe(III)-aerobactin	255 ± 7	2537 ± 33
100 μm ferrichrome	267 ± 6	1604 ± 21
100 μm ferrioxamine E	112 ± 5	756 ± 9
100 μm Ga(III)-dicitrate	32 ± 2	57 ± 3
100 μm Ga(III)-aerobactin	39 ± 3	39 ± 1
100 μm Ga(III)-desferrioxamine E	16 ± 4	7 ± 1

DISCUSSION

In this study, we present the first combined kinetic, inhibitory, and mechanistic analysis of a non-flavin-dependent FSR within bacterial systems. We discovered a pattern of association between ferric siderophore uptake systems and putative reductases with C-terminal Fe/S cluster-binding motif in several species of the Firmicutes group, in which the presence of substrate-specific FSRs was not shown before. One of these reductases, FchR, was characterized in B. halodurans and found to possess similar and distinct features to the E. coli FhuF reductase. Although utilization of an Fe/S cofactor and a substrate spectrum that includes hydroxamate siderophores was also described for FhuF, the differences to the FhuF-type reductases are significant. FchR was found to be a cytosolic enzyme and was not localized in cytoplasmic membrane fractions in contrast to E. coli FhuF (25, 26). The primary function of FhuF was found to be associated with ferric hydroxamate reduction, especially of coprogen, ferrichrome, and ferrioxamine B (26). In contrast, FchR covers a broader range of ferric siderophores, including citrate-hydroxamates and carboxylates. Furthermore, an fchR mutant was found to be growth-limited during iron deprivation even in presence of different ferric siderophores, whereas a fhuF mutant was only found to be affected during growth with ferrioxamine B as sole iron source (57, 58). In both fhuF mutant (26) and fchR mutant, intracellular accumulation of iron in the presence of different ferric siderophores indicates that these reductases are key enzymes for further cytosolic metabolization of these iron sources. However, different routes of iron release via the hydrolytic triscatecholate-trilactone pathway or the reductive hydroxamate pathways may form bypasses for each other in E. coli, whereas B. halodurans obviously lacks hydrolytic release pathways and hence fchR seems to be a general bottleneck for siderophore-dependent iron release processes. The close association of fchR with a ferric citrate-hydroxamate uptake system further differentiates it from lone-standing fhuf-type reductases in Enterobacteriaceae and points to its coordinated expression during iron limitation together with ferric substrate uptake genes. Interestingly, the substrate spectrum of FchR includes both carboxylate- and (citrate)-hydroxamate-type siderophores, and there was a clear relationship between redox potential of the enzyme Fe/S cofactor and the reductive potential of the ferric substrate. Thus, although translationally coupled with a ferric citrate-hydroxamate uptake system specific for Fe(III)-schizokinen and Fe(III)-aerobactin, the catalytic efficiency for Fe(III)-dicitrate reduction was higher than for ferric citrate-hydroxamates. This suggests the possibility of effective substrate competition dependent on the available iron source under native conditions. Although utilization of citrate as an iron source is a rather ubiquitous feature due to its general presence in primary metabolism, the utilization and availability of secondary metabolite siderophores vary among habitats. Thus, in a species that primarily uses ferric siderophore reduction for iron assimilation in an alkaline environment that renders citrate into a higher affinity siderophore than under neutral or acidic conditions, higher efficiency for Fe(III)-dicitrate conversion (in addition to its still favorable redox potential for fast iron release) may have been forced evolutionarily.

The production of schizokinen as a citrate-hydroxamate-type siderophore in B. halodurans is first shown in this study, and the putative schizokinen biosynthesis gene cluster that additionally contains a predicted major facilitator superfamily-type efflux transporter was identified as well. Together with the substrate specificity of FchR and the uptake-associated binding protein BH1037, a complete siderophore pathway for endogenous schizokinen utilization can now be suggested, which in B. halodurans includes seven genes (BH2618–BH2624) for biosynthesis and efflux and four genes (BH1037–BH1040) for uptake and cytosolic iron release (Fig. 1, B and D).

The observed binding of different ferric substrates with moderate to low affinities by FchR indicates a general relaxed recognition of ferric siderophore complexes leading to promiscuity within the tested substrate spectrum. In contrast, binding of the apo-forms of these siderophores was not observed, indicating that the formation of the iron-oxo centers in the siderophore complexes were crucial for establishing protein interaction. Although structural information is not present for FchR-type reductases, ferric siderophore binding via the oxygen donor atoms involved in ferric iron coordination is a common theme in bacterial ferric siderophore binding and receptor proteins (59–64), and the iron-oxo-dependent substrate recognition mode of FchR suggests that binding of this structural motif is essential during catalysis. Furthermore, K_m values were found to correspond to K_D values of protein-substrate interaction especially for substrates that have lower turnover rates. These low potential substrates show saturation of catalysis at low concentrations, likely because of longer interaction between substrate and protein according to enzyme cofactor-limited electron transfer rates. This further indicates that first-order rate constant k₂ is generally much lower than k₋₁ and further decreases with the drop of substrate redox potential, again pointing to the interdependence of cofactor redox potential and substrate binding affinity during the reductive process.

This leads to the definition of two different kinetic modes established for substrates over a wide range of the redox potential scale as follows: high K_m and k_cat values for high redox potential substrates like Fe(III)-dicitrate and low K_m and k_cat values for low potential substrates within the group of hydroxamates. The latter mode is possibly achieved by efficient binding at the ferric-oxo center until electron transfer is completed and fast product dissociation occurred. The mechanism of increasing binding affinities toward lower potential substrates likely explains the capability of electron transfer to those substrates ranging outside the effective Fe/S cluster redox potential such as ferrioxamine E which is about 70 mV below the determined midpoint range of −348 ± 59 mV. However, strong limitations of transfer apparently occur if substrate redox potentials range far below as in the case of ferric triscatecholates, which were not observed to saturate catalysis at high concentrations and which showed marginal turnover rates. Furthermore, the mechanism of electron transfer seems not to involve a ternary complex formation between Fe/S cofactor, siderophore metal center, and siderophore donor atoms. Either the resulting species is kinetically too unstable that it cannot be trapped by redox-inert metal centers or, more likely, charge transfer does not proceed directly via cofactor interaction with the siderophore metal center. This unlikely inner sphere electron transfer has been reported so far only for metal chelate reduction by small reducing molecules or free metal ions (7). Still, an inner sphere electron transfer via a catalytically active residue that transiently occupies the substrate metal coordination sphere cannot be excluded. However, because kinetic exchange rates between ferric ions and their siderophore ligands are generally rather low (65), electron transfer via an outer sphere mechanism is more likely in terms of fast reaction kinetics and could involve the intrinsic ligand-to-metal charge-transfer, which is mediated by the ligand donor atoms.

Despite the capability of reducing substrates, including low potential ferric hydroxamates, the strong enhancement of catalytic rates in the presence of an Fe(II) acceptor during the reaction suggests that iron release can be a rate-limiting step during iron assimilation if ferrous iron scavengers are either saturated or, on the other hand, not present in sufficient amounts. However, as shown previously, induction of iron-cofactor-binding proteins such as B. subtilis or E. coli suf genes (66, 67) occurs during iron-limiting conditions, which could contribute to an increase of iron release rates by cytosolic iron sequestration. Thus, the enhancement of ferrous iron release observed in the presence of apo-SufU, which was most likely due to equilibrium displacements between free substrate (Fe(III)·L) and product (Fe(II)·L) concentrations, points to a possibility of shifting redox potentials in the cytosol as predicted previously (18). Thus, the mechanism of released iron scavenging by efficient intracellular sinks such as the Fe/S cluster or heme assembly systems can be seen as a further mode of increasing the actual redox potentials of the FSR substrates. Because these intracellular systems are ubiquitous, this strategy can be referred to as a rather general one in addition to compartment-specific mechanisms of low potential increase like extracellular or periplasmic acidification or the intracellular drain of iron during extracellular release that is coupled with efficient uptake such as in the yeast FRE/FTR1-FET3 system (15). Interestingly, the highest ratios of 5–6-fold enhancement of catalytic efficiencies were observed for Fe(III)-schizokinen and Fe(III)-aerobactin, not for Fe(III)-dicitrate, indicating limitations of increasing turnover rates for high potential substrates by product sequestration at least under the examined conditions in vitro. In summary of the present findings and previous knowledge, a current model of cytosolic iron release mechanisms in bacterial systems is presented (Fig. 8).

FIGURE 8.

Current model of cytosolic iron release systems in Gram-positive and Gram-negative bacteria. On the left side, the pathway for iron release by specific hydrolysis of intrinsic siderophore ester linkages in trilactone scaffolds such as ferric enterobactin (hydrolyzed by E. coli Fes) or ferric bacillibactin (hydrolyzed by B. subtilis BesA) is indicated. On the right side, specific and unspecific reactions for ferric siderophore reduction are depicted. Reaction with enzymes encoded by iron-regulated genes and specifically interacting with a certain group of ferric siderophores are represented by Fe/S cluster-dependent E. coli FhuF (most specific for ferrioxamine B) and B. halodurans FchR (most specific for ferric dicitrate and ferric citrate-hydroxamates). Furthermore, unspecific possibilities of reduction (at least observed in vitro) are indicated by E. coli Fre and Fpr transferring electrons via released or stably bound flavin (Fl) cofactors, respectively. As shown in this study, iron-scavenging apoproteins such as Fe/S cluster assembly proteins and, putatively, further iron sinks such as heme and storage proteins can be regarded as (direct or indirect) acceptors of the released cytosolic iron. The intracellular binding capacity of iron may serve both as an enhancer or buffer of the upstream release reactions, mainly by influencing the actual cytosolic ferric siderophore redox potentials.

A further observation made during this study was that Ga(III)-charged siderophores can act competitively as inhibitors of reductive iron release in vitro and in cell culture. In vitro, Fe(III)-dicitrate is effectively competed by Ga(III)-desferrioxamine with a low micromolar K_i value and less effectively by Ga(III)-dicitrate, which is in agreement with the different binding constants found for the ferric complexes of these chelators. Because single turnover of Fe(III)-dicitrate in the presence of excess of Ga(III)-dicitrate occurred within several minutes as observed during CD measurements, FchR inhibition by these mimics may not only depend on direct competition with the ferric substrate at the substrate-binding site but may further interfere with the regeneration of the FchR reduction potential during multistep catalysis. Inhibition in vivo further depends on the uptake capacity for Ga(III)-loaded siderophores. Ga(III)-aerobactin and Ga(III)-dicitrate were taken up with similar efficiency, and Ga(III)-aerobactin showed the most significant inhibition of growth, whereas Ga(III)-desferrioxamine was taken up poorly and thus had no stronger growth inhibitory effect, in contrast to its stronger inhibitory effect shown in vitro. The inhibitory effect of redox-inert siderophore mimics was clearly demonstrated here for the FchR-dependent growth of the B. halodurans WT strain showing a strong contrast to the ΔfchR mutant. Interestingly, the utilization of iron source mimetics based on gallium as a pharmacologically studied and Food and Drug Administration-approved metal (68–69) has already shown to be successful during animal infection model studies using P. aeruginosa and Francisella tularensis (70, 71). Thus, the inhibition study presented here provides a further rationale for observed antibiotic effects of gallium-siderophore mimetics and may underlie further therapeutic investigations. These may include pathogens that are predicted to use reductive iron release strategies within their virulence-relevant siderophore pathways, as for example Bacillus anthracis and Bacillus cereus (72), Mycobacterium tuberculosis (73, 74), Y. enterocolitica (73), P. aeruginosa (73, 75), or Staphylococcus aureus (76).