In Table 1 some details and references for the 26 structures for which coordinates have been deposited in the Protein Data Bank are summarized. All of the systems listed have the ability to deposit mineralized iron oxyhydroxide cores and many of them have been engineered to probe certain structural or mechanistic features of the native systems. Three of the entries concern the ferritin-like proteins from Listeria innocua and Helicobacter pylori as well as the dps DNA- binding protein from Eschirichia coli, which are smaller than the ‘‘true’’ ferritins having only 12 rather than 24 protein subunits enclosing the central cavity. These proteins are used by the relevant bacteria to sequester iron especially under iron starvation conditions and have been recognized to have important health implications in a variety of infections. Since the iron uptake, storage, and release from these have not been studied in any great detail these will not be discussed further here and are included inTable 1for completeness and also with the expectation that future studies will shed light on the nature of iron storage and the mineral core which is expected to contain around 500 iron centers.
The other structures that have been deposited derive from both animal and bacterial forms.
One important aspect in which these forms differ is that the native bacterial forms from E. coli andRhodobacter capsulatus are found to have heme groups associated with them whereas horse, mouse, and frog ferritins are heme free. Binding studies show that heme groups can be bound to the animal proteins and for this reason the study to assess protoporphyrin IX binding (as a tin complex) to apoferritin from horse spleen (PDB code 1hrs) was performed. In addition, bacterio- ferritins fromE. colican be engineered to crystallize without heme (PDB code 1eum), thus acting like an animal ferritin, and providing a useful system for investigating iron reaction chemistry in a ferritin system which can be readily produced in the laboratory rather than having to be obtained commercially as is generally the case for animal ferritins. In addition, other engineered ferritins have been produced in order to test mechanistic and binding hypotheses as will be discussed later in this chapter.
Looking back at the unfolding of the ferritin story up to our current level of knowledge it can be seen how the development of analytical techniques has helped to improve our understanding of the structural and functional aspects of iron homeostasis. The progress made in protein crystal- lography over the years provides a particularly striking example. Thus, although ferritin was first isolated from mammalian spleen and liver by Laufberger2 in 1937, the crystallization of the protein from horse spleen was first reported in 1963 by Harrison.3These crystals were of loaded ferritin and only very low-resolution data at 6 A˚ could be obtained4,5on this orthorhombic form.
Subsequently it was found that better data could be obtained on the apoferritin (i.e., free of the metal core) which crystallizes in the cubic space group F432.6,7 Several structures for horse, mouse, human, and frog ferritins have now been determined and the coordinates deposited in the Protein Data Bank (see Table 18–29) in some cases on recombinant forms. So far for animal ferritins, only HoSF crystallizing in space groups other than F432 has been investigated and the
symmetry relations to the cubic form discussed. The ferritins from bacterial sources have been structurally investigated by X-ray crystallography and so far no high-resolution single-crystal studies on plants have been reported.
By way of introduction it is helpful to summarize the current level of understanding in regard to iron storage in ferritins. Many review articles concentrating on various aspects of the story have appeared.30–51 As will be seen in the following discussion, there remain many unanswered questions, perhaps requiring the development of new experiments and equipment before the picture can be completed.
The first protein structure determination on horse spleen ferritin (HoSF) performed by Harrison and co-workers6,7,37 revealed that 24 subunits, each comprising four helices, which can also be regarded as 12 dimeric subunits, connect together to form a spherical protein shell encapsulating a
Table 1 Summary of ferritin and related structures with coordinates deposited at the PDB.
PDB code Source
Resolution
(A˚) Comments References
1aew Horse
apoferritin
1.95 Cubic form. Engineered and expressed in E. coli
8
1bcf E. coli 2.9 Identification of heme binding 9
1bfr E. coli 2.94 Monoclinic form of bfr 10
1bg7 Frog 1.85 Engineered and expressed inE. coliwith mutation lys82gln and leu134pro
11
1dat Horse
apoferritin
2.05 Cubic form. Engineered and expressed in E. coli
12a 1dps E. coli 1.6 Ferritin 12-mer homologue that binds and
protects DNA. Mutation of ser164cys. 23 symmetry
13
1eum E. coli 2.05 Engineered to remove heme groups. Known as ecFTN
14
1f61 Mycobacterium
tuberculosis
2.0 Isocitrate lyase. Engineered and expressed in E. coli
15 1fha Human 2.4 Recombinant H-chain expressed inE. coli
with mutation lys86gln
16 1gwg Horse 2.01 HoSF light chain expressed inE. coli. Used
as part of a test for triiodide heavy atom phasing
17
1h96 Mouse 1.6 Mouse light chain expressed inE. coliwith thr121ala
18 1hrs Horse 2.6 HoLF cocrystallized with Sn-protoporphyrin
IX to investigate heme binding
19
1ier Horse 2.26 Cubic form of HoSF 20a
1ies Horse 2.5 Tetragonal form of HoSF 20a,b
1jgc Rhodobacter
capsulatus
2.6 M-free ferroxidase center and observation of heme groups
21 1ji4 Helicobacter
pylori
2.52 12-mer compare with 1dps and 1qgh 22 1krq Campylobacter
jejuni
2.70 Nonheme bacfer 23
1lbr Mouse 1.21 L-chain expressed inE. coli 24
1mfr Frog 2.8 Studies on diiron site in frog ‘‘M’’ ferritin 25
1qgh Listeria
innocua
2.35 12-mer compare with 1dps and 1ji4 26, 27 1rcc Frog 2.4 L-chain at pH 5.5 with glu57ala, glu58ala,
glu59ala, glu61ala, and betaine
28
1rcd Frog 2.0 L-chain at pH 5.5 plus betaine 28
1rce Frog 2.4 L-chain at pH 6.3 with glu57ala, glu58ala, glu59ala, glu61ala, and betaine
28
1rcg Frog 2.2 L-chain at pH 5.5 plus betaine 28
1rci Frog 2.0 L-chain at pH 5.5 with his25tyr and betaine 28
2fha Human 1.9 Expressed inE. coliwith lys86gln 8
a Compare the structures 1dat, 1ier, and 1ies. b A higher-resolution study on the tetragonal from is reported in Granieret al.29
spherical cavity 7.8 nm in diameter (Figure 1). In this case the protein crystallized with 432 symmetry and the arrangement of the subunits is such that threefold and fourfold channels (also a consequence of the site symmetry) lead into (or out of) the cavity. It was initially assumed that these would be the pathways for iron and water movement with the cavity acting as the iron storage site. Subsequent studies revealed that it is the threefold, hydrophilic channels in animal ferritins that are likely to allow for iron movement both into and out of the central cavity.
8.7.2.2 A Structural Model for Loaded Ferritins
Since the protein structure could not reveal the details of the contents of the cavity, a number of complementary techniques were used to establish the form of the iron storage. It is generally accepted that the iron is stored in the form of an iron(III) oxyhydroxide mineral which is structurally similar to ferrihydrite and that the cavity can hold up to 4,500 iron centers. The actual number of iron centers storedin vivodepends to some extent on the location of the protein in the organism, with more storage taking place in the liver and spleen than the brain, for example. These aspects will be discussed further in the following sections and for the present we can visualize the iron-loaded ferritin molecule as consisting of the following motifs (Figure 2):
(i) an organic shell which is made up of the 24 protein subunits,
(ii) nucleation sites on the interior of the shell to which iron(III) centers will coordinate, and (iii) an inorganic mineral core of iron(III) oxyhydroxide.
From this structural model we can see that a loaded ferritin molecule is an organic/inorganic hybrid nanoparticle. Furthermore, the inorganic core represents a nano-sized portion of a solid- state structure and is thus of interest to researchers in the area of nanoscience where the proper- ties of systems cannot be described using conventional ideas from molecular or from solid-state science. Also noteworthy is the control over the properties of the core which is exerted by the protein shell and which effectively provides a nanoreactor for the production of the iron(III) oxyhydroxide nanoparticle. Thus, the ferritin system provides insights into many aspects of nanoscience which are at the forefront of research in this field. Also, if we look back on how
Hydrophilic threefold channel
Hydrophobic fourfold channel
Subunit protein structure
Ferritin
Iron oxyhydroxide core represented by a model compound
Figure 1 The structural features of ferritin with a ribbon diagram illustrating the tertiary and quaternary structure of the protein shell viewed down the fourfold axis.
the secrets of ferritins were uncovered we find that progress here has been directly linked to progress and developments in the instrumentation available for investigating such materials. For example, as techniques such as electron microscopy have improved so has our knowledge of the details of the core structure and this will be discussed further below.
8.7.2.3 Current Knowledge on Structural and Physical Properties of Loaded Ferritins
The following description summarizes the knowledge on ferritins which has been reviewed since the early 1970s30–51and is designed to provide an introduction to the more detailed discussion on the coordination chemistry aspects which appears inSection 8.7.4.
Although the overall structural features of ferritins are fairly clear, some of the details are harder to resolve. This concerns principally the metallo components. Firstly and as stated above, it has not proved possible to elucidate the structure of the core using single-crystal protein crystallography, which is largely a result of the scale of the problem.
Indeed, it is a typical nanoscale problem, since locating up to 4,500 iron centers and their associated oxide and hydroxide oxygen atoms is not really feasible, while a powder pattern for such a small portion of an iron(III) oxyhydroxide mineral (which is probably also disordered) is not going to give much information on the detailed structure either, i.e., it lies between the molecular and solid-state descriptions.
In addition to the iron core, higher-resolution structural studies revealed that there can be further metallosites in the threefold channels (see Sections 8.7.4.1and8.7.4.4.2). These dinuclear ferroxidase centers are thought to be responsible for oxidizing the FeII (which is delivered and released from transferrin) for storage in the mineral core. The exact nature of these dinuclear sites at all stages of this process in the native system is not entirely clear from the available crystal- lography. Nevertheless, the fact that these sites exist is not in dispute. However, the questions concerning the mechanisms by which iron is taken up and released prove rather harder to address.
These points will be discussed further below, but for the present we can note that studies point to the iron being transported to the ferritin by transferrin. In order for the iron to be released from the transferrin it must be reduced from iron(III) to iron(II). The iron(II) then moves down the threefold channels where it encounters the ferroxidase center and is oxidized to iron(III). It then enters the cavity where hydrolysis reactions lead to the formation of the iron(III) oxyhydroxide mineral. At this point, the question has been posed as to whether all of the iron(III) within the core has been oxidized by the ferroxidase centers or whether it is only those which attach to the nucleation sites and then the rest are oxidized by the driving force of the mineral formation.
The idea that it is only the relatively small number needed for nucleation that are oxidized at the ferroxidase centers arises from the observation that oxidizing each and every iron center at the ferroxidase site could not only represent a bottleneck in the system (catalytic oxidation at the formation of an iron(III) oxyhydroxide mineral is well known in mineralogy), but also that the quantities of hydrogen peroxide produced in the animal systems, which is believed to be a
Inorganic core
Nucleation sites
Organic shell
Figure 2 A schematic representation of the structural elements of loaded ferritins.
product of the oxidation process, should be kept to a minimum. Additionally, the possibilities for creating free radicals via Fenton-type chemistry should be minimized. The difficulty here lies in following the system in experiments that represent what is occurring in vivo. This is, of course, true of most mechanistic studies on metalloenzymes.
Much less is known about iron release from the ferritin mineral core in vivo. Experiments performed in vitroshow that the iron(III) core can be removed by dialyzing with a reducing or chelating agent. Subsequently in these in vitro experiments it is then possible to lay down new cores of mineralized species, showing the utility of the apoprotein to act as a nanoreactor as mentioned above. These results are discussed further inSection 8.7.4.For the present we note that the favored mechanism for iron release involves reducing the iron(III) to iron(II).
Thus, overall, the details of core formation are not entirely clear, the exact nature of the core is still open to speculation, and the details of core dissolution need much more study.
First the techniques used to probe the structures and chemistry of the various aspects of ferritins will be described inSection 8.7.3.Then the coordination chemistry aspects in regard to the questions posed above will be discussed inSection 8.7.4.