Solution and solid state NMR approaches to draw iron pathways in the ferritin nanocage

Ferritins are intracellular proteins that can store thousands of iron(III) ions as a solid mineral. These structures autoassemble from four-helix bundle subunits to form a hollow sphere and are a prototypical example of protein nanocages. The protein acts as a reservoir, encapsulating iron as ferric oxide in its central cavity in a nontoxic and bioavailable form. Scientists have long known the structural details of the protein shell, owing to very high resolution X-ray structures of the apoform. However, the atomic level mechanism governing the multistep biomineralization process remained largely elusive. Through analysis of the chemical behavior of ferritin mutants, chemists have found the role of some residues in key reaction steps. Using Mössbauer and XAS, they have identified some di-iron intermediates of the catalytic reaction trapped by rapid freeze quench. However, structural information about the iron interaction sites remains scarce. The entire process is governed by a number of specific, but weak, interactions between the protein shell and the iron species moving across the cage. While this situation may constitute a major problem for crystallography, NMR spectroscopy represents an optimal tool to detect and characterize transient species involving soluble proteins. Regardless, NMR analysis of the 480 kDa ferritin represents a real challenge. Our interest in ferritin chemistry inspired us to use an original combination of solution and solid state approaches. While the highly symmetric structure of the homo-24-mer frog ferritin greatly simplifies the spectra, the large protein size hinders the efficient coherence transfer in solution, thus preventing the sequence specific assignments. In contrast, extensive (13)C-spin diffusion makes the solution (13)C-(13)C NOESY experiment our gold standard to monitor protein side chains both in the apoprotein alone and in its interaction with paramagnetic iron species, inducing line broadening on the resonances of nearby residues. We could retrieve the structural information embedded in the (13)C-(13)C NOESY due to a partial sequence specific assignment of protein backbone and side chains we obtained from solid state MAS NMR of ferritin microcrystals. We used the 59 assigned amino acids (∼33% of the total) as probes to locate paramagnetic ferric species in the protein cage. Through this approach, we could identify ferric dimers at the ferroxidase site and on their pathway towards the nanocage. Comparison with existing data on bacterioferritins and bacterial ferritins, as well as with eukaryotic ferritins loaded with various nonfunctional divalent ions, allowed us to reinterpret the available information. The resulting picture of the ferroxidase site is slightly different with various ferritins but is designed to provide multiple and generally weak iron ligands. The latter assist binding of two incoming iron(II) ions in two proximal positions to facilitate coupling with oxygen. Subsequent oxidation is accompanied by a decrease in the metal-metal distance (consistent with XAS/Mössbauer) and in the number of protein residues involved in metal coordination, facilitating the release of products as di-iron clusters under the effect of new incoming iron(II) ions.