Iron Detoxification Properties of Escherichia coliBacterioferritin

Bacterioferritin (EcBFR) ofEscherichia coli is an iron-mineralizing hemoprotein composed of 24 identical subunits, each containing a dinuclear metal-binding site known as the “ferroxidase center.” The chemistry of Fe(II) binding and oxidation and Fe(III) hydrolysis using H2O2 as oxidant was studied by electrode oximetry, pH-stat, UV-visible spectrophotometry, and electron paramagnetic resonance spin trapping experiments. Absorption spectroscopy data demonstrate the oxidation of two Fe(II) per H2O2 at the ferroxidase center, thus avoiding hydroxyl radical production via Fenton chemistry. The oxidation reaction with H2O2 corresponds to [Fe(II)2-P]Z + H2O2→ [Fe(III)2O-P]Z + H2O, where [Fe(II)2-P]Z represents a diferrous ferroxidase center complex of the protein P with net charge Z and [Fe(III)2O-P]Z a μ-oxo-bridged diferric ferroxidase complex. The mineralization reaction is given by 2Fe2+ + H2O2 + 2H2O → 2FeOOH(core) + 4H+, where two Fe(II) are again oxidized by one H2O2. Hydrogen peroxide is shown to be an intermediate product of dioxygen reduction when O2 is used as the oxidant in both the ferroxidation and mineralization reactions. Most of the H2O2produced from O2 is rapidly consumed in a subsequent ferroxidase reaction with Fe(II) to produce H2O. EPR spin trapping experiments show that the presence of EcBFR greatly attenuates the production of hydroxyl radical during Fe(II) oxidation by H2O2, consistent with the ability of the bacterioferritin to facilitate the pairwise oxidation of Fe(II) by H2O2, thus avoiding odd electron reduction products of oxygen and therefore oxidative damage to the protein and cellular components through oxygen radical chemistry.

The ferritins are a family of iron storage proteins that reversibly sequester iron in a mineral form that can be mobilized for intracellular metabolism. Ferritins from Escherichia coli are of two types: the non-heme bacterial ferritins EcFtnA 1 and EcFtnB, and the heme-containing bacterioferritin EcBFR. Although EcFtnA is a true ferritin with an iron storage role, the function of EcBFR is less clear (1). Both types of ferritin are composed of 24 identical subunits each containing ferroxidase and nucleation sites (1)(2)(3)(4)(5). These sites are important for rapid oxidation of Fe(II) to Fe(III) and its subsequent hydrolysis to form a mineral core on the inner surface of the protein. In the E. coli bacterioferritin, up to 12 protoporphyrin IX heme groups of unknown function bind between 2-fold symmetry-related subunits and are ligated by methionines Met-52 and Met-52Ј. Although the hemes do not possess catalase activity, they may be involved in mediating iron-core reduction and iron release from the protein (1)(2)(3)6).
The three-dimensional structures of ferritins from different origins are highly conserved despite homology as low as 15% in amino acid sequence in some cases. The typical 24-mer ferritins assemble in highly stable complexes with 4/3/2 symmetry, forming a large central cavity of about 80 Å diameter which can accommodate up to 4500 iron atoms as a ferric hydrous oxide mineral core. Whereas ferritins from various organisms share many common structural features, their chemistries of iron deposition differ significantly (1, 2, 6 -11). Such differences are undoubtedly important for the physiological function of each protein in iron storage and detoxification. Most notably, hydrogen peroxide is the product of dioxygen reduction in the ferroxidase reaction of mammalian ferritins, whereas water is produced in EcBFR and in the 12-mer ferritin from Listeria innocua (LiFtn) (6,9,11).
To account for the complete reduction of O 2 to H 2 O in EcBFR, two ferroxidation mechanisms have been proposed (6), one involving a high valent iron oxidation state (Fe 4ϩ ) similar to that of the di-iron enzymes ribonucleotide reductase and methane monooxygenase (12,13), and the other involving the reduction of O 2 to H 2 O 2 at one di-iron ferroxidase center followed by rapid reduction of H 2 O 2 to H 2 O at a second di-iron ferroxidase center. The previous observation that H 2 O 2 is a better oxidant for Fe(II) in bacterioferritin than O 2 (6) is consistent with the latter mechanism and has implications for the role of the protein in avoiding Fenton chemistry inherent in the odd electron oxidation of iron(II) by H 2

MATERIALS AND METHODS
The purification of EcBFR from cell extracts obtained from an overexpression system was achieved by methods described previously (14) but using an overexpression system different from the earlier one, i.e. a pET vector instead of the previous pALTER. Plasmid pGS758 (6), which contained the gene for E. coli bacterioferritin, was digested with NdeI/ EcoRI, and the fragment containing the bfr gene was ligated into pET21a (Novagen) digested with NdeI/EcoRI. The resulting plasmid was used to transform E. coli strain BL21 (DE3). 250-ml cultures were grown in LB medium containing 100 g/ml ampicillin to an A 650 nm of 0.8 -0.9 and then induced with 1 mM isopropyl-␤-D-thiogalactopyranoside (final concentration). After induction for 3 h, cells were harvested, washed, and stored at Ϫ70°C until use. The heme content of the purified EcBFR was ϳ2.5 heme per 24-mer.
Non-heme iron was removed from EcBFR by reduction with sodium dithionite and complexation with 2,2Ј-bipyridyl (15). The protein concentration was determined spectrophotometrically using an absorptivity of 33,000 cm Ϫ1 M Ϫ1 at 280 nm for the apoprotein (6) or by the Advanced Protein Assay (www.cytoskelton.com, patent pending). Freshly prepared stock solutions of H 2 O 2 were assayed by the amount of O 2 produced upon addition of catalase as measured by Clark-type electrode oximetry or from its absorbance at 240 nm (⑀ ϭ 43.6 cm Ϫ1 M Ϫ1 ). The enzyme catalase (EC 1.11.1.6, 65,000 units/mg) was purchased from Roche Molecular Biochemicals, the Amplex Red hydrogen peroxide assay kit from Molecular Probes (Eugene, OR), and EMPO from Oxis Research (Portland, OR). All other compounds including Mes and Mops buffers, sodium chloride, and ferrous sulfate were of reagent grade quality.
The use and standardization reactions of the oxygen electrode/pHstat apparatus are described in detail elsewhere (6,9). For the anaerobic experiments, all solutions were thoroughly deoxygenated using high purity grade argon gas (99.995%, Ͻ5 ppm O 2 ). The evolution of O 2 after the addition of catalase to H 2 O 2 containing EcBFR solutions or O 2 consumption during Fe(II) oxidation in EcBFR was followed either at room temperature (22°C) or at 25°C by a Clark-type oxygen microelectrode. Typical conditions for measurements were 1 M protein in 100 mM NaCl and 0.3 mM Mes, pH 6.5 to 7.0 (controlled by the pH-stat), with increments of 24 -48 Fe(II)/protein, added as freshly prepared ferrous sulfate solution at pH 3.5 (J. T. Baker Inc.). The addition of 0.3 mM Mes buffer to the protein solution improved the stability of the pH-stat control without significantly buffering the solution. Background corrections for the free acid in the ferrous sulfate solutions were made in all calculations. Conditions for each experiment are given in the figure legends.
The ultraviolet-visible difference spectrophotometric titrations of Fe(II) oxidation by H 2 O 2 in EcBFR were performed on a Cary 50 spectrophotometer. For the anaerobic experiments, 1 ml of deoxygenated apo-EcBFR solution was added to an argon-purged quartz cuvette with a micro stir bar at the bottom. A constant positive atmosphere of pure argon was maintained through the rubber septum of the cuvette during the experiment to avoid oxygen diffusion into the cell. An apoprotein solution in the same buffer was used as a blank for absorbance measurements. The experiments were terminated when no further increase in the absorbance was seen following the addition of Fe(II) to the protein solution.
EPR measurements were carried out on a laboratory assembled EPR spectrophotometer described elsewhere (16). In the EMPO spin trapping experiments for hydroxyl radical (17), the spectra were recorded immediately after the addition of the last reagent. The conditions of the experiment and the parameters of the spectrometer are indicated in the figure legends.
In the Amplex fluorescence assay for H 2 O 2 , 50 mM Mops buffer, pH 7.4, was used in place of the 50 mM phosphate buffer, pH 7.4, supplied with the assay kit in order to avoid possible interference by the phosphate on the iron chemistry in ferritin. Fluorescence standard curves were constructed using duplicate samples and measured at the same time as the unknowns which were both incubated for 30 min at room temperature and protected from light. Fluorescence intensity was measured on a Varian Cary Eclipse fluorimeter using excitation and emission wavelengths of 560 and 590 nm, respectively. The detection limit of the Amplex H 2 O 2 fluorescence assay and the standard deviation in the fluorescence intensity measurements are reported elsewhere (18).

RESULTS
Previous studies of iron uptake by EcBFR using O 2 as oxidant have shown that Fe(II) binding at the dinuclear centers of the protein followed by its rapid oxidation and core formation occur in three distinct phases (6,19). We describe the stoichiometric equations and the kinetics of iron oxidation and hydrolysis in EcBFR using H 2 O 2 as the oxidant. We demonstrate for the first time that H 2 O 2 is an intermediate product of iron oxidation by O 2 in EcBFR and report that bacterioferritin minimizes the production of highly reactive hydroxyl radicals generated by the Fenton reaction (Equation 1) facilitating pairwise oxidation of Fe(II) by H 2 O 2 .

Fe(II) Binding and Oxidation in EcBFR by H 2 O 2 -
The number of protons produced upon Fe(II) binding to the protein followed by oxidation with H 2 O 2 was monitored by titration with standard NaOH (5 mM) to maintain the pH at 7.0 with the pH-stat apparatus. The anaerobic addition of 24 Fe(II) to the apoprotein in the absence of H 2 O 2 resulted in the production of two protons per Fe(II) added (first and second injections, Fig.  1), consistent with previous results obtained with EcBFR at pH 6.5 (6). 2 The binding reaction for this first step is thus written as shown in Equation 2, where [Fe(II) 2 -P] Z represents a diFe(II)-protein complex at the ferroxidase center.
To determine the number of protons produced during Fe hydrogen peroxide solution, which was prepared at pH 3.5 in deoxygenated water. In another experiment, either 24 or 48 Fe(II)/shell were added anaerobically to the protein solution followed immediately by the addition of H 2 O 2 . In both cases, no H ϩ was produced.
The anaerobic addition of Fe(II) to EcBFR showed no UV absorbance, but upon the addition of H 2 O 2 to the protein solution containing 48 Fe(II)/shell, a band with a maximum absorbance at 295 nm emerged (Fig. 2, inset). The spectrometric titration shown in Fig. 2 has a discontinuity in absorbance at 0.5 H 2 O 2 /Fe(II), indicating that one H 2 O 2 oxidizes two Fe(II). 3 Because two Fe(II) are oxidized by one H 2 O 2 and no H ϩ is generated upon H 2 O 2 addition, we write Equation 3 for the oxidation reaction as follows: where [Fe(III) 2 O-P] Z represents a -oxo-bridged species at the dinuclear center of the protein.
The absorptivity for the observed oxidation product [Fe(III) 2 O-P] Z is 3460 M Ϫ1 cm Ϫ1 per iron at 295 nm, which is very similar to the value of 3380 M Ϫ1 cm Ϫ1 observed for EcBFR when O 2 was the oxidant (6). Similar spectrometric titrations have been reported for human H-chain and horse spleen ferritins with absorptivities of 2990 and 3540 M Ϫ1 cm Ϫ1 , respectively (9). Absorbances within this spectral region of ferritins and other dinuclear Fe(III)-proteins generally arise from -oxobridged Fe(III) dimer species (6, 9, 20 -22). Therefore, we attribute the absorbing species seen in Fig. 2 to a -oxo-bridged dinuclear complex formed at the ferroxidase center of the protein. This assignment is also based on data from a preliminary Mössbauer study with EcBFR (23) and spectrometric titration data for EcBFR using O 2 as the Fe(II) oxidant (6).
In order to determine the Fe(II)/protein stoichiometry for Fe(II) oxidation, a careful anaerobic spectrometric titration of EcBFR with Fe(II) followed by addition of H 2 O 2 was carried out. Excess H 2 O 2 was added to different samples of apo-EcBFR previously loaded with various amounts of Fe(II) (see Fig. 3 legend). A discontinuity in absorbance at 48 Fe(II)/protein (Fig.  3) is evident, consistent with the binding and oxidation of two iron at each of the 24 ferroxidase centers of the protein. Similar UV titrations have been reported (6,9) for the ferroxidation reactions of EcBFR and HuHF with O 2 as the oxidant. The result in Fig. 3  was also determined when Fe(II) was added beyond the 48 Fe(II)/shell required for saturation of the ferroxidase sites. The anaerobic addition of 500 Fe(II)/protein followed by H 2 O 2 titration resulted in an increase in the absorbance intensity (Fig. 5, inset), giving a ratio of two Fe(II)/H 2 O 2 (Fig.  5). The absorptivity of the mineralized iron was determined to 3 The increase in absorbance beyond the 48 Fe(II)/protein ratio seen in Figs. 2 and 5 was due to the slight excess of H 2 O 2 needed to fully oxidize all the iron added and does not indicate a two-step oxidation process. The ratio of H 2 O 2 /Fe 2ϩ was determined from the intercept of the two computed tangents. 4 where FeOOH (core) is a mineral core with UV absorption properties similar to that observed previously (6,9) for bacterioferritins and mammalian ferritins when O 2 was used as the oxidant. Hydroxyl Radical Spin Trapping-The oxidation stoichiometry of two Fe(II)/H 2 O 2 (Fig. 2) implies that Fenton chemistry should minimally occur in EcBFR. EPR spin trapping experiments were therefore undertaken to determine whether hydroxyl radical production is attenuated by the presence of the protein. All solutions were thoroughly deoxygenated with argon before each experiment. A control reaction corresponding to the addition sequence, DTPA ϩ EMPO ϩ H 2 O 2 ϩ Fe(II) in Mops buffer, pH 7.4, was recorded (Fig. 7, spectrum A). The strong EPR signal seen in spectrum A reflects the efficiency of the 1 electron oxidation of the Fe(II)-DTPA complex in generating a hydroxyl radical through the Fenton reaction (17). Spectrum B was obtained in another control experiment carried out in the absence of DTPA for the addition sequence, EMPO ϩ H 2 O 2 ϩ Fe(II), and has an intensity of 68% of that when DTPA is present. A third control experiment showed no EPR signal when EMPO and H 2 O 2 were added to the apoprotein alone (spectrum C). For the addition sequence, EcBFR ϩ EMPO ϩ H 2 O 2 ϩ Fe(II), a signal corresponding to 21% of that of spectrum A was produced (spectrum D). A much weaker EPR signal (8% of spectrum A) was obtained when EcBFR was first loaded with 48 Fe(II)/shell in the presence of EMPO, followed by the addition of H 2 O 2 (spectrum E). Thus, detectable hydroxyl radical production during Fe(II) oxidation by H 2 O 2 is significantly attenuated by the protein.

Fluorescence Assay for H 2 O 2 Production-Experiments
To test the possibility that hydroxyl radicals are produced during the aerobic addition of Fe(II) to apo-EcBFR, the addition sequence EcBFR ϩ EMPO ϩ 48 Fe(II)/shell was carried out. An EPR signal identical to that of spectrum E (spectrum F, Fig. 7) was again obtained corresponding to 8% of spectrum A. Therefore, when O 2 is used as the Fe(II) oxidant, a small amount of the H 2 O 2 produced at the ferroxidase center subsequently reacts with iron via the Fenton reaction.
Two other samples were prepared, 200 Fe(II)/shell (spectrum G , Fig. 7) and 500 Fe(II)/shell (spectrum H, Fig. 7) in order to determine whether H 2 O 2 addition to a preformed Fe(III) mineral core would generate hydroxyl radicals. In these samples, Fe(II) was added to the apoprotein in the presence of O 2 and continuously stirred for 6 -8 h to ensure complete oxidation and followed by the addition sequence EMPO ϩ H 2 O 2 . A very weak EPR signal (2.8% of spectrum A) was obtained in both cases and indicates that a small amount of hydroxyl radical of unknown origin is produced. DISCUSSION Ferritins have developed a variety of mechanisms to counter the problem that iron imposes as a facilitator of oxygen radical chemistry. The present work shows that EcBFR utilizes H 2 O 2 to oxidize iron pairwise, thus preventing the formation of hydroxyl radical through Fenton chemistry. A stoichiometric ratio of 0.5 H 2 O 2 per Fe(II) is obtained from spectrophotometric data for both ferroxidation (Fig. 2) and mineralization (Fig. 5) reactions, indicating that EcBFR is capable of nullifying the harmful combination of Fe(II) and H 2 O 2 . A similar stoichiometry was obtained using the oximetry apparatus, where the effect of the enzyme catalase on the remaining H 2 O 2 in solution, after its reaction with Fe(II), indicates pairwise oxidation of Fe(II) by H 2 O 2 (Fig. 4). These data are consistent with the fact that traditional ferritins have an iron housekeeping role, making iron less likely to participate in free radical-generating reactions (1, 10).
The EPR spin trapping experiments showing little HO ⅐ radical production are in agreement with the two Fe(II)/H 2 O 2 stoichiometry and indicate a dependence of the amount of hydroxyl radical trapped on the order of addition of reagents (Fig.  7). Namely, when Fe(II) is first bound to the protein followed by the addition of H 2 O 2 (spectrum E, Fig. 7), only 8% of the hydroxyl radical is generated compared with the control experiment (spectrum A, Fig. 7). A stronger EPR signal is obtained (21% of control) when H 2 O 2 is first added to the apoprotein followed by Fe(II) (spectrum D, Fig. 7), presumably because of competition between Fe(II) binding at the ferroxidase center and the Fenton reaction. Nevertheless, it is clear that Fenton chemistry is significantly reduced in EcBFR regardless of the order of addition of H 2 O 2 and Fe(II), but reduction of HO ⅐ radical formation is most pronounced if Fe(II) is pre-bound to the protein prior to introduction of H 2 O 2 .
The fluorescence assay experiments reported here indicate that H 2 O 2 is an intermediate product of dioxygen reduction when O 2 is used as the oxidant in both the ferroxidation and mineralization reactions. The present results support the twostep ferroxidation mechanism where H 2 O 2 is produced at one ferroxidase center and is rapidly consumed at the other to produce H 2 O (see Introduction). The 10-fold more rapid reaction of H 2 O 2 with the di-Fe(II) ferroxidase site compared with O 2 (see "Results") is consistent with this mechanism. Because the two protons generated per Fe(II) are from ferrous binding to the apoprotein, we write the mechanism of iron oxidation by O 2 as Equations 6 -12.

Diferrous binding,
Step one, [Fe(II) 2  behavior was seen with mammalian ferritins where H 2 O 2 produced at the ferroxidase center of these proteins accumulates to measurable levels in the bulk solution and can be detected by the addition of catalase (9, 24 -26). In accord with these observations, experiments with HuHF (not shown) indicate that O 2 is the preferred oxidant for Fe(II) over H 2 O 2 , the opposite of what is found for EcBFR. Because small amounts of H 2 O 2 were detected by the fluorescence assay in EcBFR (see "Results"), and previous catalase experiments had no effect on the iron oxidation rate and on the Fe(II)/O 2 stoichiometry (6), H 2 O 2 must become quickly consumed by other Fe(II) bound to the ferroxidase centers as shown in the reaction mechanism. It is evident, however, that a small amount of Fe(II) reacts via Fenton chemistry (Fig. 7).
One might anticipate the formation of a peroxo-diFe(III) intermediate during Equation 7 as found for other ferritins when O 2 is the oxidant (27)(28)(29). To date, there has been no indication of a peroxodiferric intermediate formation in EcBFR; more studies of the iron-complex intermediates formed in the early and late stages of its ferroxidase reaction are needed. The fact that no peroxo intermediate has been observed with bullfrog H-chain protein, although such an intermediate is seen in other ferritins (27)(28)(29), does not exclude its formation as it may be formed in low amounts, being rapidly consumed in a subsequent reaction as suggested by Pereira et al. (30). Initially, the peroxo complex was difficult to see in EcFtnA because of its low absorbance but is more evident in site-directed variants of this protein (21,27).
That the chemistry examined in this work takes place at the 24 ferroxidase centers of EcBFR is implied by the stoichiometry of 48 Fe(II)/shell when H 2 O 2 is used as oxidant (Fig. 3). The fact that the molar absorptivities and absorption maxima of the ferroxidase center iron and the mineral core in EcBFR when H 2 O 2 is used as the oxidant are essentially the same as those when O 2 is used as the oxidant (6) suggests the formation of the same -oxo or -hydroxyl species within the protein independent of the oxidant employed.
Bacterioferritin (EcBFR) and bacterial ferritins (EcFtnA and EcFtnB) are only distantly related in evolution (1, 7) and appear to have different primary functions. Whereas EcFtnA is primarily expressed during the growth phase, in keeping with a general housekeeping role in iron metabolism, EcBFR is a stationary phase protein consistent with an involvement in stress relief (1). Further evidence for a role for bacterioferritin in the response to oxidative stress comes from the work of Ma et al. (32) who showed that Pseudomonas aeruginosa bacterioferritin A is required for protection against H 2 O 2 , and Chen and Morse (33) who demonstrated that Neisseria gonorrhoeae bacterioferritin protected the organism from iron-mediated oxidative stress. The demonstration that the bacterioferritin comigratory protein (34), which is induced by oxidative stress, is a thiol-peroxidase (35) indicates that the cell has multiple means of removing H 2 O 2 when it is oxidatively stressed under iron-rich conditions. As noted earlier, both the ferroxidase center and the mineralized core of EcBFR consume H 2 O 2 (Figs. 2, 3, and 5), consistent with this ferritin being important for relief from free radical toxicity within the cell. In this context the H 2 O 2 reaction might be as significant as an iron storage function of bacterioferritins, with the Fe(II)-protein complex acting as a peroxidative co-factor in a similar manner to the thiol and heme groups of other peroxidases. In this connection, a recent investigation of E. coli Dps, a DNA-binding protein, showed it to have a ferritin-like capacity to bind and store iron using H 2 O 2 as oxidant, thus preventing the production of reactive oxygen species and oxidative damage to DNA (31). EcBFR may have a similar role in protecting cellular components against radical assault.
In conclusion, the ability to utilize and sequester iron safely is a fundamental requirement of nearly all living organisms, and although the precise physiological function of bacterioferritin remains to be established, a role in preventing iron-induced radical toxicity is suggested by the present study. The fate of the H 2 O 2 produced during iron deposition in ferritins is of considerable importance. EcBFR may provide a protective effect within the cytosol by oxidizing iron pairwise with H 2 O 2 , similarly to the Dps protein (31), thus avoiding Fenton chemistry.