Iron Oxidation in Escherichia coli Bacterioferritin Ferroxidase Centre, a Site Designed to React Rapidly with H2O2 but Slowly with O2

Abstract Both O2 and H2O2 can oxidize iron at the ferroxidase center (FC) of Escherichia coli bacterioferritin (EcBfr) but mechanistic details of the two reactions need clarification. UV/Vis, EPR, and Mössbauer spectroscopies have been used to follow the reactions when apo‐EcBfr, pre‐loaded anaerobically with Fe2+, was exposed to O2 or H2O2. We show that O2 binds di‐Fe2+ FC reversibly, two Fe2+ ions are oxidized in concert and a H2O2 molecule is formed and released to the solution. This peroxide molecule further oxidizes another di‐Fe2+ FC, at a rate circa 1000 faster than O2, ensuring an overall 1:4 stoichiometry of iron oxidation by O2. Initially formed Fe3+ can further react with H2O2 (producing protein bound radicals) but relaxes within seconds to an H2O2‐unreactive di‐Fe3+ form. The data obtained suggest that the primary role of EcBfr in vivo may be to detoxify H2O2 rather than sequester iron.

Abstract: Both O2 and H2O2 can oxidise iron at the ferroxidase centre (FC) of Escherichia coli bacterioferritin (EcBfr) but kinetic details of the two reactions are unclear due to H2O2 being an intermediate of iron oxidation by O2. UV-vis, EPR and Mössbauer spectroscopies were applied to follow the reactions when a protocol was used in which O2 or H2O2 was added to apo-EcBfr pre-loaded anaerobically with Fe 2+ . We show that O2 binds di-Fe 2+ FC reversibly, two Fe 2+ ions are oxidised in concert and a H2O2 molecule is formed and released to solution. This peroxide molecule further oxidises another di-Fe 2+ FC, at a rate ~1000 faster than O2, ensuring an overall 1:4 stoichiometry of iron oxidation by O2. Initially formed Fe 3+ can still react with H2O2 (producing protein bound radicals) but relaxes within seconds to an H2O2-unreactive di-Fe 3+ form. The data obtained suggest that the primary role of EcBfr in vivo is rather H2O2 detoxification than iron sequestering.

Over-expression and purification of EcBfr and variants
Wild type and variant EcBfr proteins were prepared as previously described, [1] with expression induced with 25 µM IPTG. Sodium dithionite and bipyridyl were used to remove non-haem iron. [2] The non-haem iron content in the protein after the procedure has been assessed previously and found to be low (~1 iron per 24mer). [3] The assessment of selected samples in this study gave a range of iron content of 4-10 iron per 24mer or 16-40% of monomer concentration. From the titration of apo-EcBfr with Fe 2+ , it follows that most of this residual ferric iron is not in the ferroxidase centre (FC), and from the intensity of the g = 4.3 EPR signal in the apo-EcBfr (which usually corresponds to 0.5-1% of monomer concentration), it follows that most of the residual ferric iron is antiferromagnetically coupled. The monomeric concentrations of EcBfr were determined using the following ε280 values: 33 000 (WT), [4] 25 585 (Y25F), [1a] 24 600 (Y58F), [1a] 23 375 (W133F), [1b] 22 300 (W35F) [1a] all in units of M -1 cm -1 . The concentration of haem iron was determined using a ε418 value of 107 000 M -1 cm -1 . [5] after non-haem iron removal. All variants were found to contain 0.3-1.5 haem/EcBfr.

Anaerobic buffers and O2 concentration controlled solutions
A Schlenk line was used to prepare anaerobic protein and buffer solutions (100 mM MES, pH 6.5) under argon. Saturated O2 solutions were prepared by bubbling O2 through buffer; the concentrations of O2 in these solutions were calculated as the maximum O2 solubility for the room temperature and the local atmospheric pressure [6] on the day of experiment. This was typically around 1.2 mM O2.

Spectrophotometric titrations of EcBfr with H2O2 or O2
Degassed proteins in sealed cuvettes were loaded with degassed ferrous ammonium sulfate solution in 50 mM HCl and then titrated with H2O2 solution in degassed buffer or O2-saturated buffer (see above). UV-vis spectra were recorded more than 1 min after each incremental addition, when any consecutively measured spectra became identical, on a Cary UV-vis spectrophotometer (Agilent Technologies). Incremental dilution during the titration was accounted for in presenting the spectra and the analysis.

Stopped-flow UV-vis measurements
Transient kinetics of the reactions were monitored using an Applied Photophysics SX20 stopped-flow spectrometer equipped with a thermostat (used at 25 o C and 10 o C) and either a photodiode array (PDA) multi-wavelength unit or a photomultiplier. The protein solutions were prepared anaerobically on a Schlenk line before being incubated with Fe 2+ (see above). Iron loaded EcBfr and anaerobically prepared buffer with either H2O2 or O2 were transferred carefully to gas-tight syringes which were fitted to the stoppedflow apparatus. Experimental kinetic traces were analysed individually and globally using the Pro-K software (Applied Photophysics).

Slow-freeze method of making samples for EPR spectroscopy
To achieve reaction time (before freezing) from ~10 s and longer (slow-freeze), EcBfr solutions were placed in selected Wilmad SQ EPR tubes (Wilmad Glass, Buena, NJ) with OD = 4.05±0.07 mm and ID = 3.12 ± 0.04 mm (mean ± range). In making (apo-EcBfraerobic + Fe 2+ ) samples, 5 µl of FeCl2 in HCl was delivered using a long needle to the pre-dispensed 250 µl of apo-EcBfr at ambient oxygen. The tubes were then frozen in methanol kept on solid CO2. In making (apo-EcBfr + Fe 2+ )anaerobic + H2O2 samples, 250 µl of degassed EcBfr solutions, pre-loaded with iron anaerobically, were placed to the bottom of an EPR tube carefully filled beforehand with argon. Hydrogen peroxide solution was then delivered to the protein using plastic tubing, and the mixture was frozen in methanol kept on solid CO2.

EPR spectroscopy
Low temperature EPR spectra were recorded on a Bruker EMX (X-band) EPR spectrometer with the use of an Oxford Instruments liquid-helium system and a spherical high-quality ER 4122 (SP 9703) Bruker resonator. Free radical concentrations were assessed by using a 100 µM Cu 2+ concentration standard. The instrumental conditions of EPR spectra measurements were as follows, if not stated otherwise: microwave frequency νMW = 9.47 GHz, microwave power PMW = 0.05 mW, modulation frequency νm = 100 kHz, modulation amplitude Am = 3 G, time constant τ = 82 ms, scan rate V = 0.60 G/s, number of scans per spectrum NS = 1.

Anaerobic Rapid Freeze-Quenching
Rapid Freeze-Quenched (RFQ) EPR samples were prepared on an isopentane-free apparatus as described before. [7] To perform freeze-quenching of reaction mixtures under anaerobic conditions, a glove bag was attached to the apparatus, filled with continuously flowing argon and kept for ~30 min while the RFQ syringes were equilibrated with argon by pushing the plungers up and down several times to remove air. Anaerobic solutions in airtight glass syringes where then brought inside the bag, incubated there for additional 30 min and used to fill the RFQ syringes.

RFQ samples of the ((EcBfr-57 Fe 2+ ) + H2O2)anaerobic system for parallel EPR and Mössbauer spectroscopies
The 57 Fe-containing (NH4)2Fe(SO4)2 was prepared from powdered 57 Fe (Cambridge Isotope Laboratories, Inc.) as follows. 57 mg of powdered metallic 57 Fe was dissolved in 500 μl of 2 M H2SO4(aq) and 150 mg of (NH4)2SO4 dissolved in 130 μl of water. Both solutions were heated to 95 o C before combining to generate an aqueous solution of (NH4)2Fe(SO4)2. The sample was crystallised by cooling for 15 mins in a ice/water bath before isolating by vacuum filtration. The crude product was purified by re-crystallising from boiling water. The dry mass of re-crystallised product was 242 mg which gives a yield of 62% based on Fe. Ferritin activity assessed by the assays using the synthesized (NH4)2 57 Fe(SO4)2 was indistinguishable from that assessed using (NH4)2Fe(SO4)2 of 99% purity (Sigma).
A volume of anaerobic 83.3 M WT apo-EcBfr (3 ml) was mixed inside the glove bag with an aliquot of HCl solution of 57 Fe 2+ , sufficient to fill all FCs. This mixture was then loaded into one of three Ar-equilibrated syringes of the RFQ apparatus; anaerobic water and a freshly prepared solution of 4 mM H2O2 dissolved in anaerobic water were simultaneously loaded into two other syringes. Protein samples were mixed with water or H2O2 and the mixtures were sprayed specific time thereafter onto a rotating aluminium disk thermostatically equilibrated with liquid nitrogen. The longer reaction time (60 s) samples was made using a double-push regime when an aging hose long enough to accommodate a whole one shot volume (the 400 ms hose, 350 µl to fill it) was filled with the reaction mixture, then, after a delay of 59.6 s, the second push expelled the aged volume onto the cold disk of the RFQ apparatus. More than one shots, at identical RFQ settings, were performed to increase the total amount of RFQ ice to be shared between the EPR and Mössbauer samples.
The frozen mixtures were then packed into EPR tubes and Mössbauer cups for parallel assessment by the two spectroscopies.

Mössbauer spectroscopy
The Mössbauer spectra were recorded on samples contained in Delrin cups at 4.2 K on a low-field Mössbauer spectrometer equipped with a Janis SVT-400 cryostat or at ca. 5 K on a strong-field Mössbauer spectrometer equipped with an Oxford Instruments Spectromag 4000 cryostat containing an 8 T split-pair superconducting magnet. Both spectrometers were operated in a constant acceleration mode in transmission geometry. The isomer shifts were referenced against that of a room-temperature metallic iron foil. Analysis of the data was performed with a home-made program. [8]

Figure S1
Figure S1. UV-vis spectra of the EcBFR-Fe 2+ complex titrated with oxygen. Apo-EcBFR (5.13 µM) was incubated anaerobically with 246 µM Fe 2+ (which is 5.13 µM x 24 subunits per EcBFR x 2 iron binding sites per FC) and then MES buffer (100 mM, pH 6.5) saturated with O2 (1.27 mM) was titrated into the mixture. The spectrum of the apo-EcBFR is in black. After Fe 2+ was added, a small increase in the absorbance observed in the UV region indicates some iron oxidation and must be associated with residual O2 in the system (either in the apo-EcBFR, or in the Fe 2+ solution or in both) as it cannot be accounted for by the Fe 2+ contribution to the spectrum because the control spectrum of a buffer solution with 200 µM Fe 2+ , multiplied by a factor of 10 (purple trace at the bottom of the Figure), is of a very low intensity. The blue spectra represent subsequent additions of aliquots of O2-saturated buffer. The inset displays the ΔA340 values, corrected for dilution of the protein throughout the titration, as function of [O2]. The red arrow indicates the expected O2 concentration corresponding to the 4:1 stoichiometry of iron oxidation by oxygen (Fe 2+ : O2). The two data points for the [O2] = 0 correspond to the apo-EcBFR (black trace) and EcBFR loaded with Fe 2+ , before O2 buffer aliquots were added (red trace); this corresponds to 4.6 % of total iron oxidised. .             Table S2. The nuclear parameters determined within the two identical iron sites hypothesis (a single doublet δ = 1.28 mm s -1 , ΔEQ = 3.21 mm s -1 , see Table S2) are close to the average values reported for E. coli RNR (δ = 1.26 mm s -1 , ΔEQ = 3.13 mm s -1 ). [10] When the spectrum was simulated assuming two different iron sites, the parameters (δ1 = 1.24 mm s -1 , ΔEQ1 = 3.13 mm s -1 ; δ2 = 1.32 mm s -1 , ΔEQ2 = 3.27 mm s -1 , see Table S2) were similar to those obtained for the ferroxidase centre in the ferrous state of the bacterioferritin from Desulfovibrio vulgaris [11] (δ1 = 1.22 mm s -1 , ΔEQ1 = 3.20 mm s -1 , δ2 = 1.46 mm s -1 , ΔEQ2 = 3.24 mm s -1 ). Two distinct ferrous sites with similar parameters have been identified in soluble methane monooxygenase hydroxylase from Methylosinus trichosporium OB3b (δ1 = 1.26 mm s -1 , ΔEQ1 = 3.22 mm s -1 ; δ2 = 1.35 mm s -1 , ΔEQ2 = 2.37 mm s -1 . [12] Figure S10

Figure S2
Figure S10. Mössbauer spectra of BFR samples treated with H2O2 and frozen 45 ms (blue) and 60 s (red) thereafter. The spectra were measured at 6 K with a 60 mT external magnetic field applied along the γ-beam. The spectra are normalised to a common total integral assuming the total concentration of 57 Fe is the same in the two samples. Specifically, the diagram shows hypothetical water molecules and hydroxyl groups (shown in blue colour) that might participate in coordination of the oxidants and in proton exchange. O2 binds reversibly to one of the two iron ions of the di-ferrous FC, relatively rapidly but weakly as indicated by KD = 8.23 x 10 -4 M (at 25 o C, Figure 3 and Table S1). This makes the population of the 2Fe 2+ -O2, and hence the observed rate of iron oxidation by O2, low (~0.9 s 1 at 25°C and 260 µM O2, see Figure 3B). This rate, although consistent with literature values for EcBfr with low haem content, [13] is notably lower than the pseudo-first-order values reported for iron oxidation by oxygen in other di-iron enzymes, for example, 26 s -1 in ToMOHred [14] or 22 s -1 (at 700 µM [O2]) in MMOH. [15] Second order O2 binding by the diferrous centre of Hr was reported with a kon ~10 7 M -1 s -1 . [16] An estimate of the pseudo-first-order rate constant for O2 binding to Hr, calculated from the second order value of 2.6 x 10 7 M -1 s -1 [16a] and an O2 concentration of 260 µM, yields k1 = 6700 s -1 , three orders of magnitude greater than kmax observed here.
The rate of iron oxidation by H2O2, on the other hand, is comparable with those of dedicated peroxidases, and for a range of H2O2 concentrations similar to ambient O2 concentrations is ~3 orders of magnitude higher than pseudo-first-order rate constant of oxidation by O2 (see Figure 5B vs Figure 3B).
The diagram also provides tentative assignments of the five ferric sites identified by the Mössbauer spectroscopy. Site 1 -one of the two ferric atoms of the 'dissymmetrical' di-ferric µ-oxo-bridged FC. Site 2 -the other of the two ferric atoms of the 'dissymmetrical' di-ferric µ-oxo-bridged FC. Sites 1 and 2 are antiferromagnetically coupled to an overall diamagnetic di-ferric state. Site 3 -two identical ferric atoms in the 'symmetrical' µ-oxo-bridged FC, also yielding an overall diamagnetic state FC. Site 4 -two identical ferric atoms in a 'symmetrical' µhydroxo-bridged FC, also diamagnetic. Site 5 -two identical FC ferric atoms un-bridged and therefore paramagnetic. We propose that Site 5 is formed from its precursor, Site 4, by accepting a second ligand X to already formed µ-hydroxo di-ferric FC (Figure 8) which breaks the antiferromagnetic coupling and creates two paramagnetic high spin ferric species. Since the paramagnetism is not manifested as appearance of distinct two new species, it is possible that ligand X is also a hydroxide OH -.

Supporting tables
Table S1 Table S1. The maximal rate constants of iron oxidation in the FC of EcBfr by O2 and the O2 dissociation constants KD as determined at two different temperatures (with reference to Figure 3   [a] Full-width at half-maximum Table S3  Table S3. Parameters of the theoretical traces used in simulation of the Mössbauer spectra in Figure 7A and Figure 7B (colour-coordinated with Figure 7 and Figure S11).