Paired Bacillus anthracis Dps (Mini-ferritin) Have Different Reactivities with Peroxide*

Dps (DNA protection during starvation) proteins, mini-ferritins in the ferritin superfamily, catalyze Fe2+/H2O2/O2 reactions and make minerals inside protein nanocages to minimize radical oxygen-chemistry (metal/osmotic/temperature/nutrient/oxidant) and sometimes to confer virulence. Paired Dps proteins in Bacillus, rare in other bacteria, have 60% sequence identity. To explore functional differences in paired Bacilli Dps protein, we measured ferroxidase activity and DNA protection (hydroxyl radical) for Dps protein dodecamers from Bacillus anthracis (Ba) since crystal structures and iron mineralization (iron-stain) were known. The self-assembled (200 kDa) Ba Dps1 (Dlp-1) and Ba Dps2 (Dlp-2) proteins had similar Fe2+/O2 kinetics, with space for minerals of 500 iron atoms/protein, and protected DNA. The reactions with Fe2+ were novel in several ways: 1) Ba Dps2 reactions (Fe2+/H2O2) proceeded via an A650 nm intermediate, with similar rates to maxi-ferritins (Fe2+/O2), indicating a new Dps protein reaction pathway, 2) Ba Dps2 reactions (Fe2+/O2 versus Fe2+/O2 + H2O2) differed 3-fold contrasting with Escherichia coli Dps reactions, with 100-fold differences, and 3) Ba Dps1, inert in Fe2+/H2O2 catalysis, inhibited protein-independent Fe2+/H2O2 reactions. Sequence similarities between Ba Dps1 and Bacillus subtilis DpsA (Dps1), which is regulated by general stress factor (SigmaB) and Fur, and between Ba Dps2 and B. subtilis MrgA, which is regulated by H2O2 (PerR), suggest the function of Ba Dps1 is iron sequestration and the function of Ba Dps2 is H2O2 destruction, important in host/pathogen interactions. Destruction of H2O2 by Ba Dps2 proceeds via an unknown mechanism with an intermediate similar spectrally (A650 nm) and kinetically to the maxi-ferritin diferric peroxo complex.

Dps (DNA protection during starvation) proteins are a family of bacterial proteins in the ferritin superfamily, first identified in Escherichia coli as dodecameric DNA-binding proteins in cells stressed by starvation (1). The proteins are stable nanocages with large (5-nm diameter) central cavities (2,3) and have been observed in vivo as part of biocrystalline complexes with DNA (e.g. see Ref. 4). Protection of DNA from damage by free radicals, the common property of Dps proteins, is accomplished by converting oxidants and iron released during stress to benign, hydrated, ferric oxide minerals inside the Dps protein cage (2,3). Because only some Dps protein dodecamers bind DNA but all protect DNA from hydroxyl radical damage in vitro, the mechanisms of Dps protein protection of DNA from oxidant damage and other stresses such as nucleases, radiation, and heat are subjects of active study (e.g. Refs. [2][3][4][5][6][7][8][9]. When DNA-Dps interactions occur, N-or C-terminal protein extensions that protrude from the protein nanocages are thought to be involved (10,11). The ability of Helicobacter pylori, Borrelia burgdorferi, and other pathogens to survive oxidative burst killing within macrophages and neutrophils has been attributed to Dps protein dodecamer protection of DNA based on the increased oxidant sensitivity of dps deletion mutants (e.g. Refs. [12][13][14][15][16][17][18]. Bacteria with two dps genes are rare although present in many Bacilli (see Fig. 1). No comparative studies of the reaction properties of pairs of Dps proteins have been made to our knowledge.
The ferritin superfamily, of which Dps protein dodecamers are members, are protein nanocages that form protein-encased minerals of hydrated ferric oxide through a series of steps. First, oxidation of two ferrous ions occurs at catalytic sites in the protein cage that are related to catalytic sites in di-iron oxygenases, such as stearoyl-acyl carrier protein ⌬9-fatty acyl desaturase, ribonucleotide reductase, and methane monooxygenase, except that in the ferritins the iron is a substrate (3). In maxiferritins, the first step in iron mineralization is the oxidation of coupled diferrous ions with O 2 via a diferric peroxo complex (19 -23), with partly characterized decay products such as diferric hydroperoxide complex (24) to form diferric oxy mineral precursors. Then, by mechanisms unknown, the mineral precursors move through the protein cage to the cavity and are concentrated by clustered carboxylates on the inner surface of the protein cavity and, later, at the mineral surface; inorganic hydrolysis and polymerization in the ferritin cavity form the hydrated ferric oxide minerals. Bacterial Dps proteins are miniferritins with 12 subunits (2,3,25) that contrast with the 24 in eukaryotic and bacterial maxi-ferritins. The location of the catalytic site in the protein nanocage is also different as are the active site ligands and the oxidant substrate (2, 3, 22, 26 -28). dps genes occur in organisms inhabiting a wide range of environmental niches, including those comparable with primitive, terrestrial atmospheres in at least 130 bacterial and Archaea species (8,29). The stable self-assembled ferritins, 4 ␣-helix-bundle subunits that interact to create the internal cavity (diameters of 5 nm in Dps proteins and 9 nm in maxi-ferritins) for the mineral (17, 22, 25, 27, 30 -35) also have gated pores at the junctions of triple subunit sets, with highly conserved gates thought to control the entry and exit of iron through the protein shell (3,36). Structural features specific to the Dps mini-ferritins include location of the Fe 2ϩ oxidation sites between two subunits at the surface of the cavity rather than in the middle of the subunit as in maxi-ferritins (2, 17, 25, 27, 30 -32, 37, 38) and the consumption of H 2 O 2 during iron uptake, oxidation, and mineralization rather than the release of H 2 O 2 as in eukaryotic ferritins (22,26). In the E. coli and Listeria innocua Dps protein dodecamers, H 2 O 2 is the preferred oxidant based on ferroxidase rates (26 -28).
The two dps genes, common in Bacillus spp., have different regulatory properties (6,15,39,40) and, as explored in this report, functional differences. MrgA in B. subtilis, for example, now known to be a dps gene, is regulated by PerR and responds to peroxide-specific transcriptional regulation. DpsA, the second dps gene in B. subtilis, is regulated by the SigmaB factor in stationery phase and Fur in log phase, responding to glucose starvation, metal, heat, and osmotic stresses. To compare biochemical properties of Dps protein pairs, we chose to study the Dps protein dodecamers from Bacillus anthracis, since they are the only Bacillus Dps proteins for which the crystal structures are known (31) (crystal structures for two Dps proteins from Lactococcus lactis, which have no recognizable Fe 2ϩ oxidation sites, have been solved recently (41)). Based on sequence similarities to Dps proteins in other Bacilli (see Fig. 1

EXPERIMENTAL PROCEDURES
Cloning-To clone the dps genes from B. anthracis dlp-1 (Ba dps1) and dlp-2 (Ba dps2), coding regions for Ba Dps1 and Ba Dps2 were amplified from the chromosomal DNA of B. anthracis (Sterne 34-F2) by PCR with the following primers: Ba Dps1F, 5Ј-CAAACACATGAACAAACAAGTAATC; Ba Dps1R, 5Ј-CAACTGGATCCTTATTGATTCAAGGAT-CCAAC; Ba Dps2F, 5Ј-CAAACACATGGAGTACGAAAA-CAAATG; Ba Dps2R, 5Ј-TTTAAGAACGCACTTAGCAT-GGATCCAAC. NdeI and BamH1 sites (underlined) were created during amplification for subcloning into an E. coli expression vector. Amplified PCR products were cloned into Topo vectors using the TA cloning kit (Invitrogen) to create plasmids Topo II-Ba Dps1 and Topo II-Ba Dps2. The DNA fragments of Ba dps1 and Ba dps2 were then excised from Topo II-Ba Dps1 and Topo II-Ba Dps2 by restriction digestion with NdeI and BamH1 and ligated into the multiple cloning sites of the protein expression vector pET-3a (Invitrogen), yielding plasmids pET-3a-Ba Dps1 and pET-3a-Ba Dps2. Both plasmids were sequenced with T7 primer to confirm that the open reading frame contained the correct sequences.
Expression, Purification, and Properties of Recombinant Ba Dps1 and Ba Dps2-Protein expression used cultures of E. coli competent cells BL21 (DE3)-Plyss strain (Novae) and conditions previously described (37,42,43). The expression vectors were pET-3a-Ba Dps1 or pET-3a-Ba Dps2, and washed cells were stored at Ϫ20°C overnight. After sanitation, cell extracts were clarified by centrifugation for 2 h at 4°C at 43,000 ϫ g, and ammonium sulfate was added to the supernatant solution to a concentration of 60%, conditions where Ba Dps1 and Ba Dps2 remained soluble. Buffer exchange and dialysis against two changes of 25 mM Bis-trispropane (pH 7.5) was followed by ion-exchange chromatography (Mono Q from GE Healthcare) (buffer, 25 mM Bis-trispropane (pH 7.5); linear gradient, 0 to 1 M NaCl). Fractions containing Ba Dps1 or Ba Dps2, identified by SDS-PAGE (15% acrylamide), were pooled, concentrated, and applied to a Sephacryl S300 (GE Healthcare) gel-filtration column. Purified Ba Dps1 and Ba Dps2 proteins dialyzed against 100 mM Mops and 100 mM NaCl (pH 7.0) and stored at 4°C appeared to be stable for up to 2 weeks for iron-free proteins and 2 months for mineralized proteins.
Protein concentrations were determined with the Bradford Assay (Bio-Rad), and the endogenous or mineralized iron content of the purified, recombinant proteins was analyzed with a colorimetric assay for the detection of Fe 2ϩ -phenanthroline at 510 nm (44). Analysis of protein purity used electrophoresis in SDS denaturing gels (12%) and in native (4 -15%) polyacrylamide gels calibrated with the BroadRange protein marker from Bio-Rad and stained with Coomassie Blue. Proteins with apparent sizes of 200 kDa were observed by native gel electrophoresis, gel filtration, and mass spectrometry (see below) and are consistent with the dodecameric assemblies of Dps subunits. Protein purity was Ͼ95%, and the endogenous iron content was Ͻ0.1 iron atoms/protein nanocage, as previously observed (31).
To confirm the identity of the expressed protein, peptides produced by denaturation and trypsin digestion of recombinant B. anthracis Ba Dps1 and Ba Dps2 (Trypsin Gold, mass spectrophotometer grade, Promega Inc., at a 1:100 molar ratio (pH 7.8) with incubation for 18 h at 37°C, before adding 0.5 g of phenylmethylsulfonyl fluoride) were analyzed by MALDI-TOF (Bruker Biflex III, in the negative reflector mode, N 2 laser) and calibrated with external standards (Bruker Daltonics). Data were analyzed with SpectroTYPER (sequenom) software; the error of the observed masses was estimated to be 0.1%. Predicted sites of tryptic digestions and masses were obtained with Pepcutter (ExPAcy) software. Four Ba Dps1 tryptic peptides had the following masses: residues [42][43][44][45][46][47][48][49][50][51][52][53][54][55][56][57][58][59]2153 Formation of the Iron Mineral-Determination of the number of iron atoms that can be accommodated inside the Dps dodecameric protein cages used solutions of recombinant Ba Dps1 and Ba Dps2 proteins in a buffer of 200 mM Mops (pH 7.5) with 200 mM NaCl. Protein solutions were mixed with an equal volume of a freshly prepared solution of FeSO 4 in 1 mM HCl to minimize spontaneous oxidation and hydrolysis; 50-fold increments of Fe 2ϩ atoms/protein dodecameric nanocage were added up to 1000 iron atoms. The Fe 2ϩ and protein mixture was incubated for 1 h at room temperature followed by incubation overnight at 4°C to complete mineralization (conversion of ferric mineral precursors to mineral), as previously described for maxi-ferritins (36). After the incubation, Fe 2ϩ /protein mixtures were passed through G-25 columns to trap insoluble iron hydroxide precipitates on the top of the gel that were only observed at Ͼ500 Fe 2ϩ /protein dodecamers. The iron content of mineralized Dps proteins was determined by incubating with dithiothreitol (5 mM) and 2,2Ј-bipyridyl (5 mM) for 24 h at 4°C; the Fe 2ϩ -bipyridyl complex was analyzed spectrophotometrically (molar extinction coefficient: 8300 M Ϫ1 cm Ϫ1 at 522 nm).
Fe 2ϩ Oxidation Kinetics-To measure the kinetics of iron oxidation by O 2 , solutions of ferrous substrate (0.12-4.8 mM FeSO 4 in 1 mM HCl) were rapidly mixed (1 ms) with an equal volume of Ba Dps1 or Ba Dps2 protein dodecamers (2 mg/ml, 10 M) in 200 mM Mops/200 mM NaCl at pH 7.0 aerobically or anaerobically using a stopped-flow, UV-visible spectrophotometer (*, Applied Photophysics, Surrey, UK); the molar ratio of Fe 2ϩ :protein ranged from 2 to 480 iron atoms per Dps protein dodecamer (1-40 iron per subunit) as previously described (26,37). Data were collected at 350 nm (e.g. Refs. 3, 4, 25, 32, and 45), nonspecific absorbance of ferric oxy species, and at 650 nm, the absorbance maximum of the diferric peroxo reaction intermediate observed in maxi-ferritin ferroxidase reactions (19,36,43). Kinetic studies with H 2 O 2 as the oxidant for Fe 2ϩ under aerobic and anaerobic conditions used a small volume of H 2 O 2 added to protein solutions (2 mg/ml (10 M) protein dodecamers) to a final H 2 O 2 of 10 mM followed by rapid mixing with various [Fe 2ϩ ] solutions at room temperature (20°C). In a few experiments anaerobic mixtures of protein dodecamers (10 M) and Fe 2ϩ were mixed with anaerobic solutions of H 2 O 2 to initiate the reaction, but the change in the order of addition had no effect. The concentration of 10 mM H 2 O 2 , chosen to ensure complete Fe 2ϩ oxidation, is consistent with previous studies (46). O 2 was removed by purging all solutions with argon under intermittent vacuum for 10 min as described in Mabrouk et al. (47). In the absence of protein and O 2 , a control sample of Fe 2ϩ showed no oxidation (change in absorbance) during the course of the experiments.
Initial rates of Fe 2ϩ oxidation (ferroxidase activity) were calculated by fitting the linear portion of the ascending part of each kinetic trace (with R Ն 0.98) over a range of iron:protein dodecamer ratios obtained by varying the iron concentration. The initial rates at 480 Fe 2ϩ /Dps were taken as V max for Fe 2ϩ oxidation analyzed at 350 or 650 nm. All data presented were computed from measurements on at least three independent protein preparations, each analyzed three times, as previously described (37).
Effects of Ba Dps1 and Ba Dps2 on DNA Degradation by Fe 2ϩ ϩ H 2 O 2 -Supercoiled plasmid DNA from pBR322 or pUC18 was purified with the Qiaprep Midi kit (Qiagen, Chatsworth, CA) followed by ethanol precipitation. Samples for supercoiled and relaxed pBR322 were obtained from Topogen. Effects of Ba Dps1 and Ba Dps2 on DNA mobility during electrophoresis and resistance to hydroxyl radical degradation used methods similar to those in Martinez and Kolter (48). Briefly, to analyze Dps protein protection of DNA from hydroxyl radical degradation, 4 nM DNA and Dps protein were incubated at 37°C for 10 min followed by the sequential addition of Fe 2ϩ (50 M final concentration), H 2 O 2 (10 mM final concentration) with incubation for an additional 15 min at 37°C. DNA integrity was examined with 1% agarose gels in TAE buffer (40 mM Tris acetate, 1 mM EDTA (pH 8.0)), calibrated with pBR322 (supercoiled, linear (NdeI digest), and relaxed (topoisomerase digest) obtained from Topogen (Port Orange, FL)). Agarose gels were stained with ethidium bromide and were analyzed with an Alpha-imager (Alpha Innotech Corp.).

Sequence Comparisons of Bacillus Dps
Pairs-Pairs of translated dps genes in each Bacillus spp. have 60 -62% sequence identity (Fig. 1), which is similar to the sequence similarities between the animal H and L ferritin subunits that have different catalytic activities but superimposable secondary and quaternary structures (49). An array of known, translated Bacillus Dps sequences are aligned in Fig. 1 by decreasing similarity to Ba Dps1 using ClustalW. The sequence similarity for B. anthracis, Bacillus cereus, and Bacillus thuringiensis Dps is 99% and parallels the close genomic relationship of the species (50); each species has a distinct environmental niche.
Ferroxidase activity of Ba Dps1 and Ba Dps2 protein dodecamers with O 2 as the oxidant was significant and contrasted with the E. coli Dps protein previously studied (26,28). The Fe 2ϩ oxidation rates in the two Ba Dps protein dodecamers were ϳ20% that of Fe 2ϩ oxidation in 24 subunit maxi-ferritins (Table 1) assuming 1 active site/subunit in both maxi-and mini-ferritins. The rates of Fe 2ϩ oxidation by O 2 for Ba miniferritins were, by contrast, 50-and 65-fold faster than that of catalytically inactive L maxi-ferritin and E. coli Dps protein dodecamers (26,37). The rate of oxidation in both Ba Dps proteins changed at approximately 24 Fe 2ϩ /protein dodecamer, indicating that all 12 putative di-iron ferroxidase sites had been filled (Fig. 2, C and D). Such a rate increase in Fe 2ϩ oxidation above 24 Fe 2ϩ /protein dodecamer suggests either an increased Dps ferroxidase turnover rate caused by the phase transition to mineral, to Fe 2ϩ oxidation at other types of sites such as the mineral surface, or both.
The capacity of maxi-ferritin nanocages to accumulate iron as solid ferric-oxo mineral in the cavities is in the range of thousands of atoms (2,45,51). To determine the "storage" capacity of Ba Dps1 and Ba Dps2, Fe 2ϩ was added to the recombinant proteins in air at stoichiometries of 100, 200, 300, 400, 500, 600, 750, and 1000 Fe 2ϩ /protein nanocage. The maximum amount of iron that could be added to either Ba Dps1 or Ba Dps2 was 500 iron/protein, as previously observed for other Dps proteins (2,26).
Differences between B. anthracis Dps1 and Dps2 in the Reactivity with Fe 2ϩ and H 2 O 2 -There was a qualitative difference between Ba Dps1 and Ba Dps2 in the reactions with Fe 2ϩ ϩ H 2 O 2 (Fig. 3, B and C). Ba Dps1 had no detectable, protein-dependent Fe 2ϩ /H 2 O 2 reaction under anaerobic or aerobic conditions, in contrast to Ba Dps2 (Fig. 3). However, Ba Dps1 inhibited the solution reaction between Fe 2ϩ and H 2 O 2 (Fig. 3, A and  B) (note that when Fe 2ϩ was added to H 2 O 2 without protein, the results as measured here were the same aerobically or anaerobically, but only the aerobic reaction is shown in Fig. 3). Fe 2ϩ oxidation was so fast that the reaction was apparently complete during the mixing time (dead time 1 ms), and the A 350 nm , was constant during the reaction period but increased with increasing concentrations of Fe 2ϩ (Fig. 3A). When Ba Dps1 protein dodecamers were present there was also no change in the A 350 nm absorbance after mixing, but unlike the reaction of Fe 2ϩ and H 2 O 2 without protein, the A 350 nm was low and independent of the Fe 2ϩ concentration (Fig. 3B) (Fig. 3C)

TABLE 1 Ferritin kinetics in maxi-ferritin (frog) and mini-ferritins (B. anthracis)
Solutions of protein (2 mg/ml/10 M protein dodecamers in 200 mM Mops, 200 mM NaCl (pH 7.0)) and an equal volume of a freshly prepared solution of FeSO 4 (0.12-4.8 mM) in 0.001 M HCl were mixed within 1 ms at 20°C. H 2 O 2 , when used as a final concentration, was 5 mM; changing the order of addition had no effect on the results. Removal of air was achieved by purging all solutions under argon under intermittent vacuum for 10 min as described (47) and confirmed by the stability of Fe 2ϩ solutions incubated without protein for the course of the experiments.

Ferritin Fe 2؉ oxidation (A 350 nm ) a DFP (diferric peroxo) formation (A 650 nm )
Hill  (37). Note that H 2 O 2 is a product of the catalytic reaction in animal maxi-ferritins (22,57). Data were collected from 5-14 independent protein preparations and are presented as the average with the error as the S.D.
To explore the possibility that O 2 influenced the Fe 2ϩ /H 2 O 2 reaction in Ba Dps proteins, we monitored the Fe 2ϩ oxidation rates in the absence of oxygen; stability of the control Fe 2ϩ / protein mixture over the course of the measurements confirmed the absence of oxygen. Removing air had no effect on Ba Dps1. However, for Ba Dps2, the reaction rate for Fe 2ϩ with H 2 O 2 anaerobically was higher than with O 2 with or without H 2 O 2 and the same as in maxi-ferritins with O 2 when corrected for the difference in the number of subunits (Table 1). Although determining the mechanism for the O 2 inhibition of the Ba Dps2 Fe 2ϩ /H 2 O 2 reaction is beyond the scope of this investigation, a possible explanation is competition for the active site Fe 2ϩ or a side reaction with an intermediate.

Formation of the A 650 nm Reaction Intermediate (diferric peroxo) by B. anthracis with Fe 2ϩ and H 2 O 2 in the Absence of
Air-Fe 2ϩ /O 2 reactions in animal maxi-ferritins proceed via a well characterized diferric peroxo intermediate, the first reaction product detected after the addition of Fe 2ϩ (19 -23, 52), with an absorbance maximum at 650 nm (19,53). In the case of Dps protein dodecamers, no evidence of the diferric peroxo intermediate has been previously reported to our knowledge. Nor was the characteristic blue, A 650 nm -absorbing species observed in this study in the presence of O 2. However, with H 2 O 2 in the absence of O 2 , the formation and decay of a species absorbing at 650 nm was clear (Fig. 4, Table 1). The molar extinction coefficient is 1987 Ϯ 283 M Ϫ1 ⅐cm Ϫ1 for the A 650 nm species in Ba Dps2, assuming it is a diferric peroxo species, as it is in maxi-ferritins, where the extinction coefficient is approximately half that in mini-ferritins. The progress curves at 350 and 650 nm for the Fe 2ϩ ϩ H 2 O 2 reaction of Ba Dps2 protein dodecamer in the absence of air were a parallel for the first 100 ms (Fig. 4) when the A 650 nm absorbing species began to decay. As the reaction proceeded the rate of Fe 3ϩ formation measured at 350 nm decreased, as shown in the progress curve, but the absorbance continued to increase as the diferric peroxo (A 650 nm ) decayed, indicating the conversion of the diferric peroxo complex to a Fe 3ϩ O product, which has similar but not identical spectral properties to the diferric peroxo precursor at 350 nm.
The formation and decay of the diferric peroxo reaction intermediate (A 650 nm ) in the Fe 2ϩ /H 2 O 2 reaction permitted computation of kinetic parameters impossible from the absorbance of Fe 3ϩ O at 350 nm because of the spectral similarities among intermediates, products, mineral precursors, and mineral in UV-visible spectroscopy that obscures the decay of intermediates (Fig. 4). The kinetic parameters are similar to those for vertebrate maxi-ferritin kinetics (23,37,53,54), where DFP, diferric oxo/hydroxo products/mineral precursors, and ferric oxy mineral have been differentiated by UV-visible (A 650 nm ), Mőssbauer, resonance Raman, and extended x-ray absorption fine structure spectroscopies (19 -21). The k cat for the A 650 nm species in Ba Dps2 was 638 Ϯ 118 mol of Fe 3ϩ -O-O-Fe 3ϩ formed/mol of Dps2/s; the approximate t1 ⁄2 for formation was 54 ms and for decay was 1.5 s. A K app of 0.067 mM for the DFP in Ba Dps2 in the absence of air is larger than the K d for Fe 2ϩ binding in the L. innocua Dps (10 Ϫ7 M, determined by isothermal calorimetry) (28) in the absence of air.
The Hill coefficient for formation of the diferric peroxo complex in Ba Dps2 reaction was 0.75 Ϯ 0.02 (Table 1), contrasting with the positive cooperativity observed in maxi-ferritins, (37,45,55,56). When the initial rates were plotted versus Fe 2ϩ / molecule, the sharp transition at 2.0 Fe 2ϩ /site characteristic of maxi-ferritin di-iron sites was absent.
Ba Dps Protection of DNA from Fe 2ϩ /H 2 O 2 (Hydroxyl Radical Degradation)-The protection of DNA from degradation by hydroxyl radical is analyzed as changes in DNA mobility during electrophoresis in agarose gels in air using supercoiled plasmid DNA of defined sizes (e.g. Refs. 10, 26 -28, 30, and 48). In the absence of protein the well known sensitivity of pBR322 supercoiled DNA to radical damage is illustrated in Fig. 5 (Fig. 5, lane 8). With Ba Dps1, DNA decreased (Fig. 5, lane 7), indicating partial protection. Similar results were obtained with pUC18 plasmid DNA. The difference in DNA protection by the two Ba Dps proteins is likely due to the fact that Ba Dps1 will consume Fe 2ϩ and Ba Dps2 will consume both H 2 O 2 and Fe 2ϩ in the ferroxidase reaction (Figs.  2 and 3).  (19). The biphasic reaction path in the formation of Fe 3ϩ O species had a broad absorption, measured at A 350 nm . The progress curves at 650 and 350 nm were similar for the first 100 ms, when the A 650 species began to decay; there is no UV-visible spectral signature that distinguishes among diferric peroxo, diferric oxo, and ferric oxo mineral, explaining why the A 350 nm does not decrease.

DISCUSSION
A conserved biological function shared by all ferritin family members is suppression of cellular oxidative damage by sequestering iron inside the protein nanocages as a ferric oxy mineral. Different ferritins adopt different strategies to oxidize Fe 2ϩ and store Fe 3ϩ . In animal ferritins, Fe 2ϩ reacts with O 2 (31), whereas H 2 O 2 is either a product of the ferroxidation reaction (22,57) or a poor substrate (2,3). Bacterial maxi-ferritins use H 2 O 2 and O 2 with comparable Fe 2ϩ oxidation rates and have different ferroxidase site ligands (46). In the Dps proteins the relatively benign ferric oxy mineral inside the dodecameric protein cages is used to protect DNA from the oxidative damage of iron and O 2 or H 2 O 2 reactions; some Dps proteins bind DNA (7-9, 18, 27, 29, 41, 58 -60). No dps genes have been identified in eukaryotes either because of the DNA protection conferred by histones and the nuclear membrane and/or the wide sequence divergence among dps genes that makes gene searches difficult; there are a few reports of maxi-ferritins in the nucleus (e.g. Ref. 61). When two Dps proteins are present, regulation of each is different and involves a variety of different types of stress signals (6,15,39,40).
H 2 O 2 is generally the preferred substrate over O 2 , at least in Dps proteins from organisms that have only one dps gene (26,27). It is the reaction with H 2 O 2 that is considered the main factor in Dps protein dodecamer DNA protection and resistance to oxidant stress (2). However, we found that Ba Dps2 used O 2 or H 2 O 2 in air, with rates that differed only 3-fold (Table 1), not the 100-fold change observed in E. coli Dps protein dodecamers (26). Moreover, Ba Dps1 had no detectable ferroxidase activity with H 2 O 2 either aerobically or anaerobically (Fig.  3), although the activity with O 2 was comparable with Ba Dps2 (Fig. 2, Table 1). Nevertheless, both Ba Dps1 and Ba Dps2 protect DNA (Fig. 5). The difference in the activity with H 2 O 2 in the ferroxidase reactions of the two B. anthracis Dps proteins indicates that DNA protection by Ba Dps1 is restricted to removing Fe 2ϩ and O 2 , whereas Ba Dps2 can remove Fe 2ϩ , O 2 , and H 2 O 2 , which may explain the apparently greater DNA protection by Ba Dps2 than by Ba Dps1 (Fig. 5).
Ferroxidase reactions in Ba Dps1 and Ba Dps2 differ from the reactions in other Dps proteins or maxi-ferritins in several ways. For example, the ferroxidation rates of Ba Dps1 and Ba Dps2 with O 2 were faster than E. coli and L. innocua Dps proteins (26,28). In addition, rates of the anaerobic Fe 2ϩ /H 2 O 2 reaction in Ba Dps2 are faster than in air but equivalent to rates for maxi-ferritin in air and with the diferric peroxide reaction intermediate (A 650 nm ) of maxi-ferritins. The different Dps2 reactions of Fe 2ϩ may reflect the ability of Bacillus to grow under a variety of oxygen concentrations.
Detection of a diferric peroxo complex in the anaerobic reaction of Ba Dps2 with Fe 2ϩ and H 2 O 2 is without precedent among ferritins, as is the extinction coefficient; positive cooperativity is also absent. A stoichiometry of 1.94 Ϯ 0.11 Fe 2ϩ /site can be computed from the data but does not reflect the increase in rate when the Fe 2ϩ added is Ͼ2 Fe 2ϩ /site. The active site structure in Dps proteins, mainly inferred from metal sites observed in protein crystals, has a small number of metal ligands. Less information on iron ligands is currently available from mutagenesis studies than will be needed to understand the spectral observations reported here. Thus, a structural explanation of the source of the distinctive features of the A 650 nm species in Ba Dps 2 is premature but might include multiple aromatic residues nearby, accessibility to solvent/substrate, or proximity to the cavity (11,27,31). In a recent study that appeared online after submission of this manuscript and where the results in solution were related to oxidant protection by maxi-ferritin in cells (Ref. 62 and references therein), the overall reaction proposed for the maxi-ferritin catalytic site with H 2 O 2 as a substrate was 2 Fe 2ϩ ϩ H 2 O 2 ϩ 2H 2 O 3 2Fe(O)OH core ϩ 4H ϩ . How this reaction relates to the A 650 nm intermediate observed here with Ba Dps remains to be determined in the future.
The putative active site residues in Ba Dps1 and Ba Dps2 (metal sites in protein crystals (31)), which are potential drug targets to control pathogenic bacteria, are similar to those in other Dps proteins (see Fig. 1; for review, see Ref. 2) but have yet to be studied directly. However, the similarity of the metal sites in protein crystals suggests that the catalytic differences between Bacillus Dps, mini-ferritin proteins, and those in other bacteria are due to "second shell" residues yet to be identified. A role for second shell residues in maxiferritin ferroxidase activity has also been indicated recently (37). Different conformations and metal binding ligands at or around ferroxidase sites (including ordered H 2 O molecules), which result from variation in Dps primary sequences, may contribute to both the mechanistic and kinetic variability observed here and among other ferritins (e.g. Refs. 24, 37, 46, and 63).
Distributing Dps functions between the two Bacillus Dps proteins Ba Dps1 and Ba Dps2 in B. anthracis may relate to different regulatory signals. For example, in B. anthracis, Dps1 is most similar in sequence to B. subtilis DpsA, which is regulated by Sigma factor B and general stress, whereas B. anthracis Dps2, the protein which uses H 2 O 2 as a substrate, is most similar in sequence to MrgA in B. subtilis, which is regulated by PerR and H 2 O 2. Separation of function in paired Dps miniferritins of Bacilli could be analogous to the two animal maxiferritin H and L proteins, which have different catalytic activities and signal responses (3, 45, 64 -66) and which co-assemble in physiologically dependent ratios (3,45,51). Understanding how paired Dps proteins of Bacillus function in vivo and how the proteins selectively use O 2 and H 2 O 2 will clarify the contributions of Dps protein nanocages to DNA protection and iron/ oxidant metabolism in host/pathogen interactions as well as identify potential drug targets in bacterial infection.