Ferritin Distribution and Primary Structure
We begin with a definition of what constitutes ferritin - typically it is an oligomeric protein of 24 identical or similar subunits, each of molecular weight around 20 kD, forming a hollow protein shell with an external diameter of 12-13 nm, inside diameter of 7-8 nm, molecular weight about 500 kD, capable of storing up to 4500 iron atoms in a water-soluble, non-toxic, bioavailable form as ferric hydroxyphosphate micelles. There is one exception to this definition, namely the ferritin from the Gram-positive bacterium Listeria inocua (Bozzi et al., 1997) which, unlike classical ferritins, contains twelve subunits. If however, we consider the characteristic four-helical bundle with a dimetallic-binding site as a signature of ferritins (at least H-chain ferritins), we rapidly find ourselves in the presence not of a superfamily, but of a veritable evolutionary dynasty, as was pointed out in Chapter 2*.
Ferritins have been found in a wide range of species, and sequence data - some, as in the first ever sequence of horse spleen apoferritin (Heusterspreute and Crichton, 1981) determined by direct methods, but many now by DNa sequencing1, have been deposited for more than 70 ferritins. They vary in length from 154-185 residues per subunit. Some ferritins have N-terminal extensions which lie on the outside of the assembled shell and target the ferritin to a specific destination such as plastids in plants and yolk sac in snails (Andrews etal., 1992; Lobreaux etal., 1992). For example, pea ferritin is synthesized with an N-terminal extension of 75 residues, which is missing from the mature protein. The first part of this extension is a chloroplast-targetting sequence of 47 residues, which is lost on entry into the plastid. The second part, an extension peptide, is lost prior to assembly of the
* As Brian Hartley once memorably remarked 'the trouble with discussing evolution is that we weren't there when it happened!'
1A year after the publication of the horse-spleen sequence (essentially L-chain), we had determined the sequence of human spleen apoferritin (Wustefeld and Crichton, 1982). By careful analysis, Chantal Wustefeld had found peptide sequences corresponding to about 70 residues, that did not fit in to the principal sequence (which we now know to be L). We published this data together with the main sequence, pointing out that it was probably that of the H subunit - and the rest as they say is history. The teams of cloners and DNA sequencers published the complete H sequence a few years later - today it would take only a few weeks! We also got one amino-acid residue wrong out of 174 in the horse-spleen sequence, namely residue 193, which we identified as Leu, the cloners identified as Pro, and, in our 0.2 nm resolution structure, we confirmed as Pro (Gallois et al., 1997): sic transit gloria proteomics vs genomics - but at least we knew what the protein did!
24-subunit oligomer, and its loss is a prerequisite for assembly (Proudhon et al., 1989) Yet another ferritin-like protein, artemin, has C-terminal extensions that fill the cavity and prevent iron storage (De Graaf et al., 1990). Mammals contain two ferritin subunits of distinct amino-acid sequences, known as H and L. In Table 6.1 we have compared the amino-acid sequences of four mammalian ferritins, two L chains (horse and human) and two H chains (human and rat). The H-chain is typically longer than L-chain, by four amino-acid residues at the N-terminus and three or four residues at the C terminus (in the rest of this chapter, numbering is
Table 6.1 Amino-acid sequence alignment of four mammalian ferritins (Horse L chain, HoL; Human L chain, HuL; Human H chain, HuH; Rat H, RaH) and of one of the ferritins, FTN, and the bacterioferritin, BFR of Eschericia coli
HoL SSQIRQNYSTEVEAAVNRLVNLYLRASYTyLSLGFYFDRD
FTN MLKP-MIEKL-EQMNLe-YS-LLyQQMSAWCYSH
BFR MKGD-K-INYL-KLLGNe-V-INQyFLHARM-K..
50 60 70 80
DVALEGVCHFFRELAEeKREGAERLLKMQNQRGGRALFQD
..TF—AAA-LRRH-QeeMThMQ— FDYLTDT-NLPRINT NWG-KFNFDVFFHFFIDeMKhAD-FIFRILFLF-FFNL--
90 100 110 120
LQKPSQDEWGTTLDAMKAAIVLeKSLNQALLDLHALGSAQ
IK--DC-D-ESG-N--EC-LH-e-NV--S--E--K-ATDK IK--DR-D-ESG-N--RC-LH-e--V--S--E--K-ATDK VES-.FA-YSSLDELFQETYKHeQLIT-KINE-AHAAMTN -G-F■.NIGFDVFFMRRRDRA-FLDGAKN-REAIGYADRV
130 140 150 160
ADPHLCDFLESHFLDEEVKLIKKMGDHLTNIQRLVGSQAG
Q-YPTFN— QW.YVS-qHEEE-LFKSIIDKLSLAG..KS-
LGEYLFERLTLKHD
E-LYFIDKELSTLDTQN -QNYLQAQIREEG
HoL Hul HuH RaH FTN B F R
HoL HuL HuH RaH FTN B F R
HoL HuL HuH RaH FTN BFR
HoL HuL HuH RaH FTN BFR
H24 L0
Hela cells
Muscle
Thymus
Red blood cells
Brain Heart
Lymphocytes
Liver Spleen
Serum
H0L24
Figure 6.1 Schematic representation of human 'isoferritins' of different subunit composition. Each ferritin subunit is represented as a 'sausage' and subunits are packed in a symmetrical shell. Twelve of the 24 subunits are visible, with H-chain subunits stippled and L-chain subunits plain. Homopolymers of H-chain and L-chain subunits are at the top and bottom of the figure respectively. The sources of various ferritins are listed in the right hand column. Reprinted from Harrison and Arosio, 1996. Copyright (1996), with permission from Elsevier Science.
based on the H-subunit§). Whereas the H and L sequences, show only about 54 % identity, about 90 % of H-chain residues and 85 % of L-chain residues are identical. With two kinds of subunits for a 24-subunit molecule, one can build 25 'isoferritins' (Figure 6.1). Although no one would pretend to be able to isolate and characterize all of them, variations in the amount of the two subunits produced in different tissues would mean that for example, in human muscle heteropolymers rich in H subunits are predominant, whereas in liver and spleen the population of ferritin molecules has a much greater content of the L subunit. As we will see later, the effective storage of iron in mammalian ferritins requires contributions from both types of subunit. This seems to explain why we find heteropolymers rather than homopolymers, and that homopolymers are only found in pathological situations such as the Hereditary Hyperferritinaemia-Cataract Syndrome (HHCS) described in the next chapter, where L-chain homopolymers, devoid of iron, are found (Levi etal., 1998). As we progress along the phylogenic tree, from mammals to other vertebrates (Figure 6.2), we encounter ferritins (sometimes with as many as three chains, as for example in the tadpole) which all have amino-acid sequences much closer to mammalian H-chain than to L-chain ferritins. This is also true of the ferritins of invertebrates and of plants (roughly 50 % identical to H-chains of mammals and 40 % to L-chains). Once we reach the prokaryotes a new class of ferritins appears,
§ This is done 'a contre coeur', since the L-chain sequence was the first to be established. However, it makes little sense to refer to the amino terminus of H-chain ferritins as -1, -2, etc.
H24 L0
Hela cells
Muscle
Thymus
Red blood cells
Brain Heart
Ferritins
- Figure 6.2 Phylogenetic tree showing the evolutionary relationship between members of the ferritin-bacterioferritin-rabreyrthrin superfamily. Reprinted from Harrison et al., 1998, by courtesy of Marcel Dekker, Inc.
namely the bacterioferritins, which are haem containing; prokaryotes also contain haem-free ferritins. The samples of two representative bacterial ferritins from E. coli, one FTN, without haem and one, BFR with haem are also included in Table 6.1. The bacterial ferritins diverge even further from animal ferritins with only around 20 % identical residues. However, the two E. coli ferritins, BFR and FTN, show only 14 % identity with each other., The characteristic residues exclusive to H subunits but not to L, involved, as we will discuss later, in the primary site of iron oxidation (the ferroxidase site), Glu-27, Glu-62, His-65 and Gln-141, are almost entirely conserved in the four H-type chains in Table 6.1 (the exception is residue 141 of BFR, which is Glu as in L ferritins). These four residues are also conserved in plant ferritins, and three of the four are also conserved in rubrerythrins, another family of bacterial proteins that are related in their structure and evolution to ferritins and bacterioferritins. The rubrerythrins contain an N-terminal four-helix bundle similar to bacterioferritins with a similar diiron cluster, as well as a C-terminal rubredoxin-like domain containing an Fe-Cys4 cluster".
Finally, before passing to a discussion of the three-dimensional structure of the ferritin family, what of organisms which do not contain ferritins? With the increasing number of genome sequences now available and in the course of determination, we can only make what is at best a progress report. Recently it has become apparent that a number of archaebacteria, particularly hyperthermophiles, such as Pyrococcus furiosus, T. maritima, M. thermoautotrophicum, A.fulgidus do have ferritin-like proteins. However, the yeast Saccharomyces cerevisiae does not appear to have a ferritin, bacterioferritin or rubrerythrin sequence, nor does Mycoplasma genitalium, Streptococcus pyrogenes nor two Pyrococcus genomes (P. horikoshii and P. abyssi). Why these organisms do not appear to have members of the ferritin family may be attributable to their very low sequence similarity to other family members (we recall the case of BFR and FTN of E. coli mentioned above), or else they simply do not have ferritins. A good control for a genuine non-ferritin containing organism would be Lactobacillus sp. - which apparently has evolved in the total absence of iron.
Post a comment