A novel non-heme iron-binding ferritin related to the DNA-binding proteins of the Dps family in Listeria innocua.

A multimeric protein that behaves functionally as an authentic ferritin has been isolated from the Gram-positive bacterium Listeria innocua The purified protein has a molecular mass of about 240,000 Da and is composed of a single type of subunit (18,000 Da). L. innocua ferritin is able to oxidize and sequester about 500 iron atoms inside the protein cage. The primary structure reveals a high similarity to the DNA-binding proteins designated Dps. Among the proven ferritins, the most similar sequences are those of mammalian L chains that appear to share with L. innocua ferritin the negatively charged amino acids corresponding to the iron nucleation site. In L. innocua ferritin, an additional aspartyl residue may provide a strong complexing capacity that renders the iron oxidation and incorporation processes extremely efficient. This study provides the first experimental evidence for the existence of a non-heme bacterial ferritin that is related to Dps proteins, a finding that lends support to the recent suggestion of a common evolutionary origin of these two protein families.

The physiological requirement of iron in a nontoxic and available form is met in all living cells by ferritins, the iron storage proteins, which are tailored to accomodate large amounts of readily mobilizable iron inside the apoferritin shell. In eukaryotes, the apoferritin coat is made of 24 subunits of two different types called H and L (21,000 and 19,500 Da, respectively) that have 55% amino acid sequence identity (1,2). The H and L subunits are interchangeable in the assembled molecules. They share the same tertiary conformation consisting of a bundle of four antiparallel helices (A-D), a short helix (E), and a loop that connects helices B and C (3,4). Functionally, the H subunit is characterized by the presence of a specific site for iron oxidation in the four-helix bundle, the so-called ferroxidase center, whereas the L chain contains efficient nucleation sites that face the protein cavity and are thus able to favor iron accumulation (5,6).
In bacteria, notably in Gram-negative ones, the ferritin-like proteins shown functionally to bind inorganic iron can be classified in two different categories: the heme-b containing "bacterioferritins" (Bfr) found in microorganisms such as Escherichia coli (7,8) and Azotobacter vinelandii (9) and non-heme containing ferritins such as those expressed by Helicobacter pylori (10,11) and by the E. coli gen-165 gene (12,13). Bacterioferritins are the most studied ones. They are formed by a single subunit that has a low sequence similarity (about 20%) with eukaryotic ferritins but that is characterized by the same four-helix bundle conformation and by the presence of the seven amino acid residues that constitute the ferroxidase center of the H-type chains (5). The non-heme containing ferritinlike proteins expressed by H. pylori and the E. coli gen-165 gene have only about 30% sequence similarity with bacterioferritins, but sequence alignment and modelling studies predict the conservation of the ferroxidase center residues (14). Consistent with this prediction, the H. pylori protein has been shown to bind inorganic iron (10,11).
Recently, Evans et al. (15) identified four new putative prokaryotic ferritins using gene sequence analysis and an alignment that introduces a number of gaps in regions of the protein structure that correspond to ␣-helices. On this basis, the neutrophil-activating protein A encoded by the H. pylori NapA gene (15), the protein encoded by the metal-regulated MrgA gene of Bacillus subtilis (16,17), and proteins of unknown function in Treponema pallidum (18) and Anabaena variabilis 1 display sequence similarity with mammalian and proven bacterial ferritins and conserve all or most of the seven ferroxidase center amino acids. Evans et al. (15) also found a striking similarity between this new group of putative ferritins and the DNA-binding protein from starved cells of E. coli (20), which belongs to the Dps family, a diverse group of bacterial, stressinduced polypeptides that bind DNA. Almost at the same time, Peñ a and Bullerjahn (21), in a study of the DpsA protein of Synechococcus sp., assigned all of the putative ferritins of Evans et al. (15) to the Dps family also in view of the demonstration that the B. subtilis MrgA protein binds DNA (17). Unlike most members of the Dps family, the Synechococcus sp. DpsA protein contains heme, a peculiarity that was ascribed to the high similarity (Ͼ60%) between its C-terminal domain and the C-terminal half of heme-containing bacterioferritins (21). On the basis of these considerations, Peñ a and Bullerjahn proposed a common evolutionary origin for the Dps and bacterioferritin/ferritin superfamilies and speculated that the Dps proteins may have evolved as heme-or metal-binding proteins that later acquired a DNA-binding activity.
This intriguing situation and the lack of biochemical data on * This work was supported in part by grants from the Italian Ministero dell'Università e Ricerca Scientifica e Tecnologica and Consiglio Nazionale delle Ricerche (Grant 95.02942.14 and Progetto Strategico-Biologia Strutturale). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. This paper is dedicated to Professor Alessandro Ballio on the occasion of his 75th birthday.
The ferritins extracted from Gram-positive bacteria suggested that the metal-binding ancestor of the Dps proteins hypothesized by Peñ a and Bullerjahn (21) could perhaps be found among such bacteria. Listeria innocua was chosen as the potential source of ferritin, also in view of the importance of iron in the infection process caused in humans and animals by the pathogenic species, Listeria monocytogenes (22,23). A multimeric protein able to incorporate inorganic iron, and hence functioning as an authentic ferritin, was purified and characterized. L. innocua ferritin, like all bacterial ferritins, is formed by one type of subunit (18,000 Da) whose amino acid sequence shows a high similarity to the Dps proteins and does not appear to contain the ferroxidase center residues characteristic of eukaryotic ferritins. A further unique property of L. innocua ferritin is the molecular mass of the native polymer (240,000 Da), which is suggestive of a different mode of subunit assembly with respect to tetracosameric (24-mer) ferritins, which are all characterized by a mass of 450,000 -500,000 Da.

MATERIALS AND METHODS
Enzymes and chemicals were purchased from the following suppliers. Trypsin (code TRTPCK) was from Worthington, pepsin was from Sigma, cyanogen bromide was from Fluka Chemie AG, 4-vinylpyridin was from Aldrich-Chemie, guanidinium chloride (recrystallized from methanol) was from Merck, Darmstadt, the liquid chromatography solvents (HPLC 2 grade) were from Carlo Erba Reagenti, and sequencegrade chemicals were from Perkin-Elmer.
Bacterial Strain and Growth Conditions-L. innocua was grown routinely in brain-heart infusion (BHI) medium (Merck). A preculture of 500 ml of L. innocua, obtained after an incubation period of 14 h in BHI medium (Fe(III) 20 ppm), was used as inoculum of a 50 liter fermentation vessel. After 15 h of growth, under agitation and aeration at 37°C, the culture was centrifuged, and 110 g (wet weight) of cells were obtained.
Ferritin Purification and Characterization-A 110-g sample of packed cells was suspended in 360 ml of 50 mM Tris-HCl, pH 7.4, containing 1 mM EDTA, 0.5 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride and disrupted in a French press. Debris was removed by centrifugation (15,000 rpm for 40 min), the supernatant was heated at 70°C for 10 min and rapidly cooled. After centrifugation (15,000 rpm for 30 min), this material was treated with ammonium sulfate (80% w/v); the precipitate was collected by centrifugation (10,000 rpm for 20 min), dissolved in 5 ml of 20 mM Tris-HCl, pH 7.4, and dialyzed against the same buffer. To remove DNA, the material was treated with a solution of streptomycin sulfate (1% w/v) for 30 min at room temperature and centrifuged (15,000 rpm for 15 min). The protein was further purified by fast protein liquid chromatography (FPLC) using a Mono-Q column (Pharmacia Biotech Inc.) equilibrated with 20 mM Tris-HCl, pH 7.4, and eluted with a linear gradient of 0.1-0.3 M NaCl in the same buffer. Ferritin eluted at 0.25 M NaCl. In some preparations, gel fitration chromatography on an FPLC-G200 Sephadex column (Pharmacia) was carried out as a further purification step. The purity achieved during the various steps of the procedure was checked by SDS-polyacrylamide gel electrophoresis according to Laemmli (24) and by nondenaturing gel electrophoresis (25). Denaturing gels were stained with Coomassie Blue, and native gels were also stained for iron with Prussian Blue. The data presented refer to five different protein preparations. The yield of pure L. innocua ferritin was unaffected by omitting the heating step during the purification procedure.
Protein concentration was determined by the method of Lowry et al. (26) or by using the extinction coefficient E (1 cm, 1 mg/ml) ϭ 1.2 at 280 nm that was calculated from the absorption spectrum and the protein content determined by amino acid analysis. The iron content of the native protein was determined as the 2-2Ј bipyridyl complex at 520 nm (27). Iron was removed from ferritin by incubation for 24 h in 0.3% sodium dithionite in 50 mM MES-NaOH, at pH 6.0, containing 2-2Ј bipyridyl added to chelate spurious ferrous iron.
Iron incorporation experiments were performed using Fe(NH 4 ) 2 -(SO 4 ) 2 as iron donor. All iron solutions were prepared in Thunberg tubes, kept anaerobically under a nitrogen atmosphere, and used within a few hours from their preparation. The iron oxidation and incorporation kinetics were followed spectrophotometrically at 310 nm upon addition of different amounts of iron to apoferritin solutions equilibrated in air. The extinction coefficient of micellar iron at 310 nm was taken as 450 (1%, 1 cm), the value used for horse spleen ferritin (28). As a control, the rate of Fe(II) auto-oxidation was measured in parallel. After iron incorporation, the samples were subjected to gel electrophoresis under non-denaturing conditions on 6% polyacrylamide gels. The gels were stained for iron with potassium ferrocyanide and for protein with Coomassie Blue.
Fluorescence measurements were performed using a SPEX 2000 single photon counting spectrofluorimeter by excitation of the aromatic amino acid residues at 280 nm at 20°C in 20 mM Tris-HCl, pH 7.4. The bandwidths of the excitation and emission monochromators were 2 nm. Circular dichroism spectra were measured with a Jasco J-710 spectropolarimeter in the far and near UV regions at 20°C in 20 mM Tris-HCl, pH 7.4. The molar ellipticity (degree cm 2 dmol Ϫ1 ) in the far UV region ([] Maa ) was calculated on the basis of a mean residue molecular weight of 116 and in the near UV region ([] M ) on the basis of the polypeptide chain molecular mass (18,000 Da).
Sedimentation velocity experiments were carried out on a Beckman Instruments Optima XL-A analytical ultracentrifuge at 30,000 rpm and 10 or 20°C over the concentration range 0.2-0.5 mg/ml. The gradient of protein concentration in the cells was determined by absorption scans along the centrifugation radius at 280 and 310 nm with a step resolution of 0.003 cm. Sedimentation coefficients were evaluated with the software provided by Beckman and were reduced to s 20,w by standard procedures. The data have been transformed to a plot of g(s*) versus s*, an unnormalized differential distribution of apparent sedimentation coefficients, that is geometrically similar to the more commonly used Schlieren plot (dc/dr versus r).
The sedimentation equilibrium experiments were performed on a Beckman Optima XL-A analytical ultracentrifuge at 5,000 and 10,000 rpm and 10°C over the concentration range of 0.2-0.5 mg/ml. The data (step resolution 0.001 cm, 20 averages/scan) were analyzed with the Ideal 1 software provided by Beckman.
For electron microscopy experiments, the samples were negatively stained. One drop of L. innocua ferritin obtained by incubation with 500 iron atoms/apoferritin molecule in MOPS-NaOH, pH 7.0, was placed on a carbon film mounted on a 300-mesh copper grid, immediately removed, and stained with 2% phosphotungstic acid (pH 7.2). Micrographs were taken at ϫ 50,000 with a Zeiss EM 902 electron microscope operating at 60 kV. The measurement of the diameter of the particles was performed, on about 100 particles, by a graduated lens on micrographs with a final magnification of ϫ 300,000.
Amino Acid Sequence Analysis-The purified protein (10 g) was subjected to vapor phase hydrolysis in 6 N HCl at 110°C for 24 h. Amino acid analysis was performed with a Beckman System Gold analyzer equipped with an ion-exchange column and post-column derivatization with ninhydrin. Although the amino acid analysis revealed the absence of cysteines, the protein sample (1.7 mg) was suspended in 0.5 ml of 0.5 M Tris-HCl, pH 7.5 (containing 2 mM EDTA, 4 M guanidinium chloride, and 12 mol dithiothreitol), and incubated for 3 h at 55°C. Then, 4-vinylpyridin (90 mol/10 l) was added, and after a 10-min incubation, the protein was desalted by HPLC using a guard cartridge (C8, 4.6 ϫ 30 mm). This treatment was necessary in order to denature the protein, which is otherwise resistant to proteolytic attack. An aliquot (0.7 mg) of the denatured protein was suspended in 0.5 ml of 0.1 M ammonium bicarbonate and incubated at 37°C for 3 h after addition of 14 g of trypsin. A second aliquot of protein (0.5 mg) was dissolved in 0.2 ml of 5% (v/v) formic acid and incubated with 10 g of pepsin at 25°C for 5 min. The peptide mixtures obtained following enzymatic digestions were purified immediately after incubation with proteases and without lyophilization. The last aliquot of protein was dissolved in 0.2 ml 70% (v/v) formic acid, incubated with 5 mg CNBr for 24 h at room temperature in the dark, and lyophilized.
The peptide mixtures were purified by HPLC using a Beckman System Gold chromatographer on a macroporous reverse-phase column (Aquapore RP-300, 4.6 ϫ 250 mm, 7 m, Brownlee Labs) eluted with a linear gradient from 0 to 35% acetonitrile in 0.2% (v/v) trifluoroacetic acid at a flow rate of 1.0 ml/min. Elution of the peptides was monitored using a diode array detector (Beckman model 168) at 220 and 280 nm.
The amino acid sequence of peptide samples was determined by automated Edman degradation using an Applied Biosystems model 476A sequencer. Samples (0.1-0.5 nmol) were loaded onto polyvinylidene difluoride membranes (ProBlott, Applied Biosystems), coated with 2 l polybrene (100 mg/ml of 50% methanol) and run with a Blottcartridge using an optimized gas-phase fast program. N-terminal se-quence analysis of the protein was performed on samples (10 g) electrotransferred on ProBlott membranes after SDS-polyacrylamide gel electrophoresis (29) using a liquid-phase fast program.
Peptides were numbered retrospectively according to their location in the sequence, starting from the N terminus. Tryptic peptides were designated with T, peptic peptides with P, and CNBr peptides with B.
Structure Comparison-A search of the SwissProt Data Base, pairwise and multiple sequence alignments, as well as prediction of secondary structures were carried out with the programs FASTA, GAP, PILEUP, and PEPTIDESTRUCTURE, respectively, from the Genetic Computer Group sequence analysis software package (GCG, Version 8), using a VAX/VMS system (30).

Production and Purification of Ferritin from L. innocua-
Crude extracts of L. innocua cells show, in native gel electrophoresis, the presence of a band that is characterized by a significantly faster mobility relative to horse spleen ferritin and stains specifically for iron. On the basis of this finding, a purification protocol has been utilized that takes advantage of the high thermal stability of ferritins and includes a heating step at 70°C. The subsequent purification procedure, consisting of an ion exchange chromatography and a gel fitration step, yields about 10 mg of pure protein, as judged by gel electrophoresis in denaturing conditions, from 100 g of cells. In SDS gradient pore acrylamide gel electrophoresis, L. innocua ferritin appears to be composed by only one type of subunit with a molecular mass of about 18,000 Da. This value is similar to that of other bacterial ferritins and is lower with respect to those of the H and L mammalian ferritin polypeptide chains.
Amino Acid Sequence Determination-The complete amino acid sequence of ferritin from L. innocua is reported in Fig. 1. The subunit has 156 amino acid residues, yielding a molecular mass of 18.048 Da, which corresponds to the value obtained by electrophoretic analysis. The sequence was deduced following N-terminal sequence analysis of the protein up to residue 57 and isolation and identification of an almost complete set of tryptic peptides, which were ordered with the help of overlapping peptides produced by pepsin and cyanogen bromide cleavages. The reported sequence is in good agreement with the amino acid composition of the protein. The C-terminal extremity of the protein was identified on the basis of the evidence that the same sequence, ending with a C-terminal glutamic acid, was present after complete sequencing of peptides obtained from two different types of cleavage that are known not to occur at Glu-Xaa peptide bonds (see peptides T12 and B7).
Molecular Weight Determination and Electron Microscopy Data-In sedimentation velocity experiments, native L. innocua ferritin is characterized by the presence of a single symmetrical component with s 20,w ϭ 11 S (Fig. 2a); this value does not change upon removal of the iron, indicating that the iron content of the native protein is very low. For a spherical molecule, this sedimentation velocity yields a molecular mass of 240,000 Da using the partial specific volume calculated from the amino acid composition (V ϭ 0.734) and the nomogram of Wyman and Ingalls (32).
In order to obtain a direct determination of the molecular mass of the apoprotein, sedimentation equilibrium experiments were carried out. The data analysis yields 240,000 Ϯ 1,400 Da, a value which has to be compared with 450,000 -500,000 Da typical of mammalian ferritins (1,3).
Negative staining electron microscopy shows that L. innocua ferritin is a spherical protein shell that encloses an electrondense core just like horse spleen ferritin (Fig. 3). However, the size of the two molecules differs significantly. The dimensional analysis yields an average diameter of 12.15 nm for the horse spleen ferritin molecules and of 10.14 nm for the L. innocua ferritin ones. The corresponding volumes, assuming a spherical shape, are about 940 and 545 nm 3 , respectively.
Iron Incorporation Experiment-The iron content of the native protein is low and corresponds to 5-10 atoms/molecule (240,000 Da). The iron-free protein was obtained by reduction with sodium dithionite at low concentration (0.3% w/v).
Native L. innocua ferritin and the iron-free protein are able to oxidize and incorporate iron as indicated by electrophoretic analysis of the products obtained upon incubation with different amounts of ferrous iron in the presence of molecular oxygen. A progress curve of iron uptake by L. innocua apoferritin upon addition of 500 Fe/molecule is shown in Fig. 4; the halftime corresponds to 90 s and is similar to that of human rH apoferritin analyzed in parallel. Under these experimental conditions, all the iron added appears to be incorporated by L. innocua apoferritin. Upon addition of Ն 1,000 Fe/molecule precipitation of ferric iron outside the protein is observed.
Further evidence that iron is incorporated by L. innocua ferritin is provided by sedimentation velocity analysis of samples loaded with iron at pH 7.0. The sedimentation patterns show the presence of a single homogeneous peak characterized by a sedimentation coefficient, s 20,w ϭ 21-22 S (Fig. 2b), which is significantly higher than that characterizing the iron-free protein (s 20,w ϭ 11 S). The protein loaded with iron at pH 6.5 displays a fast sedimenting component of similar sedimentation coefficient (s 20,w ϭ 21-22 S) although about 10% apoferritin is still present (Fig. 2c).
Spectroscopic Characterization-The absorption spectrum of L. innocua apoferritin in the ultraviolet region is characterized by a broad peak centered at 279 nm with a shoulder at 292 nm due to tryptophan residues and a fine structure at lower wavelengths corresponding to the contribution of phenylalanines (Fig. 5a).
The CD spectrum (Fig. 5b) in the far UV region points to a high ␣-helical content (about 80%), similar to that of mammalian apoferritins, while distinctive differences are apparent in the near UV region. The L. innocua apoferritin spectrum displays one positive dichroic band of low intensity around 253 nm, due to the contribution of phenylalanine residues. The shoulders at 264 and 271 nm in the large negative peak that dominates the spectrum between 260 and 290 nm are likewise attributable to phenylalanines, whereas the shoulders at 275 and 281 nm are attributable to tyrosines. The shoulder at 287 nm and the well resolved negative peak at 294 nm can be ascribed to the 0 ϩ 850 cm -1 1 L b and the 0 Ϫ 0 cm -1 1 L b and 1 L a transitions of tryptophan residues, respectively. These assign-ments are consistent with the broadness of the absorption band and the lack of coincidence between the absorption and CD spectra (33).
The intrinsic fluorescence spectrum is characterized by an emission maximum at 325 nm upon excitation at 280 nm (Fig.  5c). The emission wavelength is similar to that of the human rH homopolymer but occurs at higher wavelengths than in the L-rich horse spleen apoferritin. As in the case of the rH homopolymer, it can be ascribed to tryptophan residues exposed to solvent (34,35). DISCUSSION The presence of a polymeric protein in L. innocua, which is able to bind and incorporate iron, is of great interest since it represents the first example of an authentic ferritin that bears a high sequence similarity with Dps proteins and thus lends support to the evolutionary link between ferritins and Dps proteins proposed by Peñ a and Bullerjahn (21). L. innocua ferritin is also the first characterized ferritin from a Grampositive bacterium.
Native L. innocua ferritin contains a very small amount of iron but incorporates the metal efficiently at a rate that resembles that of the human rH homopolymer. The maximum iron binding capacity corresponds to about 500 atoms and is thus much smaller than that of the mammalian molecule, which can harbor up to 4500 iron atoms. This difference, however, is consistent with the molecular mass of the L. innocua polymer, which is lower than that of the mammalian one, namely 240,000 Da as compared with 450,000 -500,000 Da. It is also consistent with the electron microscopy data of Fig. 3 which show that, in L. innocua ferritin, the protein cage and the iron core are both smaller than in horse spleen ferritin. A distinctive feature of reconstituted L. innocua ferritin is its homogeneity in iron content, a characteristic that may be related to the structural basis of the iron oxidation and incorporation processes. Thus, the iron cores obtained at pH 7.4 and 6.5 are all very similar and contain about 500 iron atoms, whereas in horse spleen ferritin, the reconstituted iron cores vary greatly in size (36).
Having established that the 240,000 Da iron-binding protein isolated from L. innocua behaves functionally as a ferritin, the first question that arises concerns its relationship with the different classes of the ferritin/bacterioferritin superfamily.
However, searching in the SwissProt Data Base with the L. innocua ferritin sequence as a probe did not retrieve any of the proven ferritin sequences but retrieved with the highest score eight protein sequences all belonging to the Dps family. The alignment of these sequences, obtained using the program PILEUP, is reported in Fig. 6. The comparison shows that identical amino acid residues are clustered in the region corresponding to positions 36 -73 (the numbering refers to the alignment reported in the figure). The residues marked with aster- isks represent the DNA-binding signature in Dps proteins. Using the program GAP, a pairwise comparison was made among the sequences reported in Fig. 6. The results are shown in Table I. The percentage identity among the various sequences ranges from 18.9 to 36.9, the metal-regulated MrgA protein from B. subtilis being the most similar to L. innocua ferritin.
The L. innocua ferritin sequence was then compared with the different classes of proven ferritins, i.e. heme and non-heme containing bacterial ferritins and mammalian ferritins. As discussed by Andrews et al. (14) and Grossman et al. (37), the sequences within each class (Fig. 7) have a high degree of identity, whereas the percentage identity among the three dif-ferent classes is low. However, the residues that constitute the ferroxidase center (indicated by asterisks in Fig. 7) are conserved in all the sequences except in the L subunits of mammalian ferritins, which lack the ferroxidase activity (5).
Attempts to align the L. innocua ferritin sequence with each of the sequences given in Fig. 7 failed to show significant similarities. In particular, there was no evidence for the conservation of all or part of the canonical ferroxidase center residues.
However, for the L mammalian ferritin chains, a stretch of five identical residues (A/YERLL, positions 66 -70 in Fig. 7) was observed. Upon alignment of this stretch, additional conserved residues spread all over the sequence became apparent (boxed residues in Fig. 7). The negatively charged amino acids at positions 49, 57, 60, 64, and 67 (corresponding to positions 44, 52, 55, 59, and 62 of the L. innocua ferritin sequence, see Fig. 1) located in the B helix of the four-helix bundle are of special interest for their possible functional implication. In mammalian L chains, all these residues protrude from the B helix into the internal cavity of the assembled molecule. Those at positions corresponding to 57 and 60 are the main sites of iron core nucleation on the basis of site-directed mutagenesis experiments although carboxylates at positions 49, 64, and 67 have been suggested to participate in the process as well (38). In L. innocua ferritin, a helical wheel projection (Fig. 8) reveals that an additional negative charge, carried by Asp-53 (position 48 of the L. innocua ferritin sequence), is present between glutamic acids 57 and 60. If one assumes that the B helices of L. innocua and horse spleen ferritin are oriented in a similar fashion in the assembled molecule, Glu-57 and -60 and Asp-53 would form, in the negative cluster, a core endowed with strong Fe(III) complexing capacity. In turn, this property is known to enhance the rate of Fe(II) oxidation by atmospheric oxygen as shown by the behavior of low molecular weight Fe(III)-complexing agents (like EDTA, nitrilotriacetate, and citrate). These compounds at neutral pH values promote oxidation of Fe(II) at rates similar to those observed for L. innocua ferritin (39). In L. innocua ferritin, therefore, Fe(II) oxidation and Fe(III) nucleation would occur within the same cluster of negatively charged residues facing the internal cavity of the molecule, a hypothesis that is consistent with the homogeneity of the iron cores produced in the reconstituted protein.
The helical wheel projection of Fig. 8 also permits other considerations that concern the relationship between ferritins and Dps proteins and are based on the assumption just used, namely that the B helix has the same orientation in the L. innocua and mammalian ferritin polymers. The amino acid residues common to L. innocua and mammalian ferritin L chains (boxed residues) that are not part of the proposed iron oxidation/incorporation cluster do not face the internal cavity of the molecule, but are in other parts of the helical surface involved in the stabilization of the tertiary and/or quaternary structure. A structural role can be assigned also to most of the eight amino acids of the Dps motif that occur in the helix FIG. 6. Amino acid sequence comparison among the proteins retrieved from the SwissProt Data Base using L. innocua ferritin as probe. The alignment was obtained using PILEUP. Fer_Lisin, ferritin from L. innocua; Dps_Ecoli, DNA-binding protein from starved cells of E. coli (20); DpsA_Synec, DNA-binding hemoprotein from Synechococcus sp. strain PCC7942 (21); Dps_Heldu, Dps protein from Hemophilus ducreyi, from residue 10 (GenBank TM accession number U16121); Mrga_Bacsu, metal-regulated protein from B. subtilis (16,17); Napa_Helpy, neutrophil-activating protein from H. pylori (15); Yd49_Haein, hypothetical protein HI1349 from Hemophilus influenzae (41); Tpf1_Trepa, antigen Tpf1 from T. pallidum (18); YIt2_Anava, hypothetical 20,200 Da low temperature-induced protein (ORF2) from A. variabilis (19). For the sake of clarity, the numbering is as in Fig. 7, where ferritin sequences are compared. Residues conserved in all sequences are in boldface type. Residues that represent the DNA-binding motif are marked with and asterisk. (indicated by asterisks in Fig. 8), which are positioned over the whole helical surface. In this connection, it may be mentioned that preliminary experiments with L. innocua ferritin did not provide evidence for DNA binding. In the Dps proteins, which are all oligomeric, the marker amino acids are expected to occupy different topological positions as in L. innocua. Only those residues on the exterior of the molecule will be involved in DNA binding while the others will serve a structural role. Last, the assembly of the L. innocua ferritin molecule deserves a comment. The combination of the molecular mass of the native polymer (240,000 Da) with that of the subunit (18,000 Da) suggests that the shell of L. innocua apoferritin is formed by 12 subunits and not by 24 as in mammalian and heme-containing bacterial apoferritins. Several arrangements of the subunits are compatible with the formation of a dodecameric protein with an internal cavity. X-ray quality crystals of L. innocua ferritin have been obtained and will provide information on this point.
In conclusion, the characterization of L. innocua ferritin, which behaves as an authentic ferritin endowed with iron oxidation/incorporation properties but whose sequence is related to those of the Dps proteins, has provided an unexpected proof for the proposed evolutionary relationship between the ferritin/bacterioferritin and the Dps superfamilies and calls for further biochemical studies on representative members of the two groups of proteins.