Ferritin proteins function to detoxify, solubilize and store cellular iron by directing the synthesis of a ferric oxyhydroxide mineral solubilized within the protein's central cavity. Here, through the application of X-ray crystallographic and kinetic methods, we report significant new insight into the mechanism of mineralization in a bacterioferritin (BFR). The structures of nonheme iron-free and di-Fe(2+) forms of BFR showed that the intrasubunit catalytic center, known as the ferroxidase center, is preformed, ready to accept Fe(2+) ions with little or no reorganization. Oxidation of the di-Fe(2+) center resulted in a di-Fe(3+) center, with bridging electron density consistent with a mu-oxo or hydro bridged species. The mu-oxo bridged di-Fe(3+) center appears to be stable, and there is no evidence that Fe(3+)species are transferred into the core from the ferroxidase center. Most significantly, the data also revealed a novel Fe(2+) binding site on the inner surface of the protein, lying approximately 10 A directly below the ferroxidase center, coordinated by only two residues, His46 and Asp50. Kinetic studies of variants containing substitutions of these residues showed that the site is functionally important. In combination, the data support a model in which the ferroxidase center functions as a true catalytic cofactor, rather than as a pore for the transfer of iron into the central cavity, as found for eukaryotic ferritins. The inner surface iron site appears to be important for the transfer of electrons, derived from Fe(2+) oxidation in the cavity, to the ferroxidase center. Bacterioferritin may represent an evolutionary link between ferritins and class II di-iron proteins not involved in iron metabolism.

The ability of iron to cycle between Fe(2+) and Fe(3+) forms has led to the evolution, in different forms, of several iron-containing protein cofactors that are essential for a wide variety of cellular processes, to the extent that virtually all cells require iron for survival and prosperity. The redox properties of iron, however, also mean that this metal is potentially highly toxic and this, coupled with the extreme insolubility of Fe(3+), presents the cell with the significant problem of how to maintain this essential metal in a safe and bioavailable form. This has been overcome through the evolution of proteins capable of reversibly storing iron in the form of a Fe(3+) mineral. For several decades the ferritins have been synonymous with the function of iron storage. Within this family are subfamilies of mammalian, plant and bacterial ferritins which are all composed of 24 subunits assembled to form an essentially spherical protein with a central cavity in which the mineral is laid down. In the past few years it has become clear that other proteins, belonging to the family of DNA-binding proteins from starved cells (the Dps family), which are oligomers of 12 subunits, and to the frataxin family, which may contain up to 48 subunits, are also able to lay down a Fe(3+) mineral core. Here we present an overview of the formation of protein-coated iron minerals, with particular emphasis on the structures of the protein coats and the mechanisms by which they promote core formation. We show on the one hand that significant mechanistic similarities exist between structurally dissimilar proteins, while on the other that relatively small structural differences between otherwise similar proteins result in quite dramatic mechanistic differences.

Bacterioferritin from Escherichia coli is able to accumulate large quantities of iron in the form of an inorganic iron(III) mineral core. Core formation in the wild-type protein and a number of ferroxidase center variants was studied to determine key features of the core formation process and, in particular, the role played by the ferroxidase center. Core formation rates were found to be iron(II)-dependent and also depended on the amount of iron already present in the core, indicating the importance of the core surface in the mineralization reaction. Core formation was also found to be pH-dependent in terms of both rate and iron-loading characteristics, occurring with maximum efficiency at pH 6.5. Even at this optimum pH, however, the effective iron capacity was approximately 2700 per molecule, i.e., well below the theoretical limit of approximately 4500, suggesting that competing oxidation/precipitation processes have a major influence on the amount of iron accumulated. Disruption of the ferroxidase center, by site-directed mutagenesis or by chemical inhibition with zinc(II), had a profound effect on core formation. Effective iron capacities were found to be linked to iron(II) oxidation rates, and in zinc(II)-inhibited wild-type and E18A bacterioferritins core formation was severely restricted. Zinc(II) was also able, even at low stoichiometries (12-60 ions/protein), to significantly inhibit further core formation in protein already containing a substantial core, indicating the importance of the ferroxidase center throughout the core formation process. A mechanism is proposed that incorporates essential roles for the core surface and the ferroxidase center. A central feature of this mechanism is that dioxygen cannot readily gain access to the core, perhaps because the channels through the bacterioferritin coat are hydrophilic and dioxygen is nonpolar.

Bacterioferritin (EcBFR) of

PrrC from Rhodobacter sphaeroides provides the signal input to a two-component signal transduction system that senses changes in oxygen tension and regulates expression of genes involved in photosynthesis (Eraso, J.M. and Kaplan, S. EMBO J. 19,[4237][4238][4239][4240][4241][4242][4243][4244][4245][4246][4247]. It is also a homologue of eukaryotic Sco proteins and each has a C-x-x-x-C-P sequence. In mitochondrial Sco proteins these cysteines appear to be essential for the biogenesis of the Cu A centre of respiratory cytochrome oxidase. Overexpression and purification of a water-soluble and monomeric form of PrrC has provided sufficient material for a chemical and spectroscopic study of the properties of the four cysteine residues of PrrC, and its ability to bind divalent cations, including copper. PrrC expressed in the cytoplasm of Escherichia coli binds Ni 2+ tightly and the data are consistent with a mononuclear metal site. Following removal of Ni 2+ and formation of renatured metal-free rPrrC (apo-PrrC), Cu 2+ could be loaded into the reduced form of PrrC to generate a protein with a distinctive UV-visible spectrum, having absorbance with a V V max of 360 nm. The copper:PrrC ratio is consistent with the presence of a mononuclear metal centre. The cysteines of metal-free PrrC oxidise in the presence of air to form two intramolecular disulfide bonds, with one pair being extremely reactive. The cysteine thiols with extreme O 2 sensitivity are involved in copper binding in reduced PrrC since the same copper-loaded protein could not be generated using oxidised PrrC. Thus, it appears that PrrC, and probably Sco proteins in general, could have both a thiol-disulfide oxidoreductase function and a copperbinding role. ß 2002 Published by Elsevier Science B.V. on behalf of the Federation of European Biochemical Societies.