Characterization of ferritin from human placenta. Implications for analysis of tissue specificity and microheterogeneity of ferritins

Mammalian ferritins can be resolved into multiple components by isoelectric focusing, and each tissue contains a characteristic subset of isoferritins. Ferritin isolated from human liver was compared to acidic ferritin isolated from mid-gestational human placenta to define a structural basis for ferritin heterogeneity. Placenta ferritin contained several major bands with isoelectric points in the range of pI = 4.7-5.0 which were more acidic than the predominant isoferritins of human liver. Ferritin from each tissue was resistant to denaturation by 10 M urea and appeared to be identical by electron microscopy. Circular dichroism measurements revealed that placenta ferritin had substantially less ordered secondary structure than liver ferritin. Both types of ferritin contained only two subunits when analyzed by electrophoresis in sodium dodecyl sulfate gels, but isoelectric focusing of dissociated subunits in urea revealed 6-7 different components. In this system, placenta ferritin was enriched in the more acidic subunits and it completely lacked the most basic subunits noted in liver ferritin; placental ferritin had no unique components. Differences in isoelectric points among assembled ferritins from these two tissues appear to result from different proportions of these acidic and basic subunits.     

Variation in the distribution of two human heart ferritin species. Isoferritin profile and subunit composition in normal and iron-overloaded subjects.

On polyacrylamide slab electrophoresis two ferritin species, termed fast and slow, were present in three control and four iron-loaded human hearts. The ratio of the fast to slow species varied between hearts and correlated with the distribution of isoferritins, which had pI values ranging from 4.8 to 5.6. The acidic isoferritins were prominent in all hearts, but one control and the four iron-loaded hearts contained, in addition, basic isoferritins. All ferritins were composed of two subunits, an acidic H type and a more basic HL type, which was the main subunit in normal liver. Hearts with the highest proportions of acidic isoferritin contained the highest proportion of the H subunit.     

Characterization of the binding of ferritin to the rat liver ferritin receptor

The binding characteristics and specificity of the rat hepatic ferritin receptor were investigated using ferritins prepared from rat liver, heart, spleen, kidney and serum, human liver and serum, guinea pig liver and horse spleen as well as ferritins enriched with respect to either H- or L-type subunit composition, prepared by chromatofocusing of rat liver ferritin on Mono-P or by reverse-phase chromatography of ferritin subunits on ProRPC 5/10. No significant difference was apparent in the binding of any of the tissue ferritins, or of ferritins of predominantly acidic or basic subunit composition. However, serum ferritin bound with a lower affinity. The effect of carbohydrate on the ferritin-receptor binding was examined by glycosidase treatment of tissue and serum ferritins. Tissue ferritin binding was unaffected, while serum ferritin binding affinity was increased to that of the tissue ferritins. Inhibition of ferritin binding by lactoferrin was not due to common carbohydrate moieties as previously suggested but was due to direct binding of lactoferrin to ferritin. Therefore, carbohydrate residues do not appear to facilitate receptor-ferritin binding, and sialic acid residues present on serum ferritin may in fact interfere with binding. The results indicate that the hepatic ferritin receptor acts preferentially to remove tissue ferritins from the circulation. The lower binding affinity of serum ferritin for the ferritin receptor explains its slower in vivo clearance relative to tissue ferritins.     

Characterization of the binding of ferritin to the rat liver ferritin receptor

The binding characteristics and specificity of the rat hepatic ferritin receptor were investigated using ferritins prepared from rat liver, heart, spleen, kidney and serum, human liver and serum, guinea pig liver and horse spleen as well as ferritins enriched with respect to either H- or L-type subunit composition, prepared by chromatofocusing of rat liver ferritin on Mono-P or by reverse-phase chromatography of ferritin subunits on ProRPC 5/10. No significant difference was apparent in the binding of any of the tissue ferritins, or of ferritins of predominantly acidic or basic subunit composition. However, serum ferritin bound with a lower affinity. The effect of carbohydrate on the ferritin-receptor binding was examined by glycosidase treatment of tissue and serum ferritins. Tissue ferritin binding was unaffected, while serum ferritin binding affinity was increased to that of the tissue ferritins. Inhibition of ferritin binding by lactoferrin was not due to common carbohydrate moieties as previously suggested but was due to direct binding of lactoferrin to ferritin. Therefore, carbohydrate residues do not appear to facilitate receptor-ferritin binding, and sialic acid residues present on serum ferritin may in fact interfere with binding. The results indicate that the hepatic ferritin receptor acts preferentially to remove tissue ferritins from the circulation. The lower binding affinity of serum ferritin for the ferritin receptor explains its slower in vivo clearance relative to tissue ferritins.     

Immunochemical characterization of human liver and heart ferritins with monoclonal antibodies

A library of 27 murine monoclonal antibodies was obtained by using human liver and heart ferritins as immunogens. The specificity of the antibodies for the two ferritins and their subunits was studied with five different methods. The antibodies elicited by the liver ferritin bound preferentially the immunogen and were specific for the L subunit. Some antibodies elicited by the heart ferritin had characteristics similar to the anti-liver antibodies, other ones bound preferentially the heart over the liver ferritin and were specific for the H subunit. Only two antibodies were able to bind both ferritins and subunits. Some anti-H and anti-L chain antibodies were used to develop and compare four types of immunoassay to quantitate isoferritins. The results indicate that heart ferritin is immunologically more heterogeneous than liver, the H and L subunits having large immunological differences with few, if any, identical epitopes; and that that the architecture of the immunoassays have a strong influence on the crossreactivity of the antibodies with the two isoferritins, probably because H and L chains are not arranged randomly in the assembled protein.     

Ferritin—a mediator of apoptosis?

Previously we have demonstrated an apoptosis inducing activity for a rat hepatocyte conditioned medium (CM) presumably mediated by acidic isoferritins. Here, we present support for this assumption since isoferritins purified from different rat hepatocyte CM significantly enhanced the frequency of apoptotic cells in primary rat hepatocytes, an effect completely inhibited by a neutralizing anti-H-ferritin antibody. The apoptosis induction appears to be related to a 43 kDa ferritin subunit contained in the isoferritins released from primary hepatocytes, presumably representing a ferritin heavy/light chain heterodimer. In addition, these isoferritins immunologically crossreact with antibodies raised against placental isoferritin p43-PLF (which also contains a 43 kDa ferritin subunit) and melanoma-derived H-chain ferritin, representing ferritin isoforms which reveal immunomodulatory properties. Furthermore, p53 and FasL are upregulated upon isoferritin treatment in a time dependent mode, and apoptosis induction can be suppressed by neutralizing anti-FasL antibodies. Proapoptotic Bid is upregulated too and translocated into mitochondria in primary hepatocytes exposed to the isoferritins purified from the CM. Finally, epidermal growth factor (EGF) and dexamethasone (DEX), which counteract proapoptotic mitochondrial signalling, almost completely abolished the proapoptotic effect of the hepatocyte derived isoferritins. In conclusion, our findings demonstrate that acidic isoferritins with homology to immunomodulatory ferritin isoforms (p43-PLF, melanoma-derived-H-chain ferritin) are released from hepatocytes in vitro, and are able to stimulate upregulation of p53 and mediate apoptosis involving Fas (CD95) signalling as well as addressing the intrinsic mitochondrial proapoptotic pathway.     

Characterization of ferritin in murine erythroleukaemia cells

Murine erythroleukaemic cells were studied to determine whether different isoferritins have different functions. The cells were labelled with radioactive iron and the pattern of isoferritins was analysed by chromatofocussing. No change was found after iron-loading the cells but after inducing erythroid differentiation with dimethyl sulphoxide (DMSO), iron was incorporated into both more basic and more acidic isoferritins. This was compared to ferritin subunit synthesis; DMSO induced the synthesis of a third, minor subunit whereas iron-loading had no effect. The fate of murine erythroleukaemic cell ferritin iron was followed after incubations in iron-deficient medium containing DMSO; some, but not all, of the ferritin iron was mobilized and used for haem synthesis, and the remaining iron was found amongst the more basic isoferritins. Finally, sequential radioactive iron labels were used to demonstrate that the movement of iron from ferritin to haem was compatible with the 'last-in-first-out' principle, but this could not be related to different isoferritins. These results show firstly that DMSO changes the pattern of isoferritins and ferritin subunits in murine erythroleukaemic cells. Secondly, iron associated with more basic isoferritins seems to be less easily mobilized for haem synthesis. These results support the concept that different isoferritins have different functions.     

Characterization of ferritin in murine erthroleukaemia cells

Murine erythroleukaemic cells were studied to determine whether different isoferritins have different functions. The cells were labelled with radioactive iron and the pattern of isoferritins was analysed by chromatofocussing. No changes was found after iron-loading the cells but after inducing erythroid differentiation with dimethyl sulphoxide (DMSO), iron was incorporated into both more basic and more acidic isoferritins. This was compared to ferritin subunit synthesis; DMSO induced the synthesis of a third, minor subunit whereas iron-loading had no effect. The fate of murine erythroleukaemic cell ferritin iron was followed after incubations in iron-deficient medium containing DMSO; some, but not all, of the ferritin iron was mobilized and used for haem synthesis, and the remaining iron was found amongst the more basic isoferritins. Finally, sequential radioactive iron labels were used to demonstrate that the movement of iron from ferritin to haem was compatible with the ‘last-in-first-out’ principle, but this could not be related to different isoferritins. These results show firstly that DMSO changes the pattern of isoferritins and ferritin subunits in murine erythroleukaemic cells. Secondly, iron associated with more basic isoferritins seems to be less easily mobilized for haem synthesis. These results support the concept that different isoferritins have different functions.     

Common epitopes in human isoferritins characterized by murine monoclonal antibodies

Three monoclonal antibodies against human liver ferritin were selected to study antigenic determinants (epitopes) of human isoferritins. These monoclonal antibodies were found to form immunoprecipitin lines with ferritin in double diffusion tests (Ouchterlony), indicating multiple epitopes on a single ferritin molecule. The antibodies revealed high species specificity as well. Monoclonal antibodies MA301 and MA311 appeared to recognize different epitopes, since they did not inhibit each other in competitive enzyme-linked immunosorbent assay (ELISA). However, MA309 recognized both epitopes for MA301 and MA311 with similar competitive inhibition. These epitopes were not detectable when ferritin was treated with 8M urea (pH 2.5) and were detectable upon reconstruction by dialysis against 2 M urea (pH 7.2), suggesting that these monoclonals recognize epitopes in the tertiary structure of the ferritin molecule. As a matter of fact, these monoclonals react preferentially with intact ferritin molecule and only negligibly with subunits. Isoelectric focusing patterns of human ferritins demonstrated that liver, spleen, placenta, and hepatoma cells (Li-7) transplanted in nude mice contained basic isoferritins, whereas HeLa cells (carcinoma), Wa cells (EB virus-transformed B cells), and Raji cells (Burkitt's lymphoma) contained acidic isoferritins. Human heart ferritin displayed a somewhat intermediate pattern between liver and HeLa ferritins. In spite of the heterogeneous population of human isoferritins, the dissociation constants (Kd) of the three monoclonal antibodies to liver, HeLa, and heart isoferritins were quite similar.     

A biochemical comparison of normal human liver and hepatocellular carcinoma ferritins.

1. The iron contents, gel migration rates and isoelectric-focusing patterns of normal liver and hepatocellular carcinoma ferritins from the same patients were compared. 2. Sucrose-density-gradient centrifugation showed that the number of iron atoms per ferritin molecule was decreased to approximately half in carcinoma tissue when compared with normal liver. 3. On electrophoresis, hepatocellular carcinoma ferritin migrates faster and is therefore more negatively charged than normal liver ferritin, thus refuting the general view that the more negatively charged a ferritin molecule the greater its iron content. 4. Comparison of tumour and normal liver ferritin subunit compositions on acid/urea/polyacrylamide gels showed hepatocellular carcinoma ferritin to contain an additional, more negatively charged, subunit to normal liver ferritin. 5. Isoelectric focusing showed that hepatocellular carcinoma tissue contains isoferritins with isoelectric points intermediate between the ranges of normal liver and normal heart isoferritins.     

Heterogeneity in tissue ferritins displayed by gel electrofocusing

1. Horse spleen ferritin and human liver ferritin were examined by gel electrofocusing under conditions that demonstrated equilibrium focusing. Both ferritins were resolved into multiple isoferritins. Both families of isoferritins were separable from one another. 2. Horse spleen ferritin was also resolved into five components by ion-exchange chromatography on DEAE-Sephadex A-50. Each of the major chromatographic fractions contained only a few of the isoferritins seen on gel electrofocusing. Each chromatographic fraction corresponded to different portions of the isoferritin profile. 3. These results indicate that the heterogeneity seen in many ferritins by gel electrofocusing represents structural heterogeneity in the ferritin population as isolated from the tissues.     

Mouse hepatoma and liver ferritins. Comparative structural studies.

Pure ferritin from male mouse liver produces a single band of monomers (RF = 0.199) with electrophoresis in polyacrylamide gels at pH 9.0. The five sub-bands within this monomeric band appear to represent charge isomers having the same molecular size. Ferritin from BH3 transplantable mouse hepatoma shows two overlapping bands of monomers (RFA = 0.208 and RFB = 0.240); further electrophoretic studies show that these bands represent two subpopulations of molecules differing both in charge and size. Sub-bands are not found in this hepatoma ferritin. The larger tumor ferritin reaches the same end migration position as all liver isoferritins on gradient gels, signifying a very similar or identical molecular size; however, the absence of sub-bands indicates that this hepatoma ferritin differs in charge from the homologous liver proteins. Liver and hepatoma ferritins both produce a single prominent subunit band corresponding to nominal molecular weights of 22 250 and 21 700, with polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate and dithiothreitol. With electrophoresis on polyacrylamide gradient slabs containing sodium dodecyl sulfate and dithiothreitol, both liver and hepatoma ferritins now reveal two subunits bands situated at identical positions. The polypeptides of these two closely spaced bands have a nominal molecular weight difference of less than 1000. Neither the hepatoma nor the liver seems to produce the ferritins found in the other tissue. Nevertheless, all these ferritins are composed of the same two types of subunits, albeit in different relative amounts. Observed distinctions in the ferritins from these normal or neoplastic cells must reflect differences in assembly and processing, as well as in the regulated expression of the same ferritin genes.     

Structural and functional relationships of human ferritin H and L chains deduced from cDNA clones

We have isolated essentially full-length cDNA clones for human ferritin H and L chains from a human liver cDNA library. This allows the first comparison of H and L nucleotide and amino acid sequences from the same species as well as ferritin L cDNA sequences from different species. We conclude that human H and L ferritins are related proteins which diverged about the time of evolution of birds and mammals. We also deduce the secondary structure of the H and L subunits and compare this with the known structure of horse spleen ferritin. We find that residues involved in subunit interaction in shell assembly are highly conserved in H and L sequences. However, we find several interesting differences in H subunits at the amino acid residues involved in iron transport and deposition. These substitutions could account for known differences in the uptake, storage, and release of iron from isoferritins of different subunit composition.     

A Novel Approach for the Synthesis of Human Heteropolymer Ferritins of Different H to L Subunit Ratios

Mammalian ferritins are predominantly heteropolymeric species consisting of 24 structurally similar, but functionally different subunit types, named H and L, that co-assemble in different proportions. Despite their discovery more than 8 decades ago, recombinant human heteropolymer ferritins have never been synthesized, owing to the lack of a good expression system. Here, we describe for the first time a unique approach that uses a novel plasmid design that enables the synthesis of these complex ferritin nanostructures. Our study reveals an original system that can be easily tuned by altering the concentrations of two inducers, allowing the synthesis of a full spectrum of heteropolymer ferritins, from H-rich to L-rich ferritins and any combinations in-between (isoferritins). The H to L subunit composition of purified ferritin heteropolymers was analyzed by SDS-PAGE and capillary gel electrophoresis, and their iron handling properties characterized by light absorption spectroscopy. Our novel approach allows future investigations of the structural and functional differences of isoferritin populations, which remain largely obscure. This is particularly exciting since a change in the ferritin H- to L-subunit ratio could potentially lead to new iron core morphologies for various applications in bio-nanotechnologies.     

Properties of human tissue isoferritins.

1. Human liver ferritin was separated by preparative isoelectric focusing into six fractions. 2. Except for the least acidic fraction the reactivity with antibody against spleen ferritin increased with rising pI, but with antibody against heart ferritin the reactivity decreased. 3. The highest iron content was found in the most acidic isoferritins and progressively decreased with rising pI. 4. Iron uptake was studied in apoferritin prepared from heart and liver ferritin fractions separated by ion-exchange chromatography. There was good correlation between the rate of iron uptake and pI. The most acidic fractions took up iron more rapidly than did the more basic ones. 5. Ferritin was prepared from heart, liver, spleen and kidney. There was little difference on isoelectric focusing between ferritin obtained from normal tissues and the corresponding iron-loaded tissues from patients who had received multiple blood transfusions. The iron-loaked heart ferritin invariably contained relatively more of the basic isoferritins. Normal and iron-overloaded heart ferritins were separated into isoferritin fractions by ion-exchange chromatography, and in each case there was a fall in iron content as the pI increased. The iron content of ferritin from the iron-overloaded heart was higher throughout than that from normal heart. 6. There is a relationship between the rate of iron uptake by apoferritin and pI, and this probably accounts for the variation in iron content of the isoferritins found in human liver and heart.     

On ferritin heterogeneity. Further evidence for heteropolymers

Tissue ferritins from the horse, rat, and human consist of multiple isoferritins some of which are common to more than one tissue in the same individual. Subunit analyses indicate that the ferritins from all three species are similarly composed of only two types of subunit with an approximate Mr of 21,000 and 19,000, designated H and L. The relative amounts of these subunits vary progressively throughout the isoferritin spectrum. Amino acid analyses and tryptic peptide maps indicate that the H and L subunits have extensive sequence homologies and that both are species-specific. Both subunits have been identified as the primary products of apoferritin synthesis in a wheat germ lysate programmed by rat liver mRNA. These results substantiate our proposal (Adelman, T. G., Arosio, P., and Drysdale, J. W. (1975) Biochem. Biophys. Res. Commun. 63, 1056-1062) that tissue ferritins are not unique homopolymers but families of hybrid molecules consisting of different proportions of two subunit types.     

Structural differences in ferritins from normal and malignant rat tissues.

Ferritins purified from several normal and malignant rat tissues were examined for amino acid composition, content of tryptic peptides, available sulfhydryl groups and subunit sizes and proportion. Ferritin extracted from adult kidney, neonatal liver and hepatic and renal tumors differed from the ferritin of adult rat liver in migration on electrophoretic gels and in antibody affinity, but did not differ among themselves. Nevertheless, they showed distinctive differences in amino acid composition and tryptic peptide content. All of them and also adult liver ferritin contained two major species of subunits differing in molecular weight. The proportions of subunits, and the available sulfhydryl groups of the intact ferritin molecules, differed among these tissue ferritins. On the basis of amino acid and peptide content, the ferritins of hepatomas and the renal tumor analyzed showec the greatest similarity but not identity. The ferritin of neonatal liver was next most similar. Kidney ferritin differed considerably in composition from tumor and neonatal ferritins, while adult liver ferritin was the most extremely divergent of the series examined. A similar progressive difference was found on examining the proportions of subunits and sulfhydryl groups in these ferritins. However, changes in subunit proportion cannot explain the amino acid and peptide compositional changes.     

Multiple subunits in human ferritins: Evidence for hybrid molecules

The subunit composition of human heart and liver ferritins was examined by both sodium dodecyl sulfate gel electrophoresis and acetic acidurea gel electrophoresis. These analyses indicated that both tissues contained two subunit types of similar size but different surface charge. One subunit was common to both tissues. The implications of these findings in relation to the known heterogeneity of isoferritins are discussed, and a new model of ferritin structure is proposed.     

Characterization of autoantibodies to ferritin in canine serum

Ferritin-binding proteins circulating in mammalian blood are thought to be involved in the clearance of ferritin. The present study characterizes canine serum autoantibodies (IgM and IgA) that react with ferritin. Canine IgM and IgA bound to bovine spleen ferritin as well as canine liver ferritin. To examine the specificity of canine IgM and IgA to ferritin H and L subunits, we used canine heart ferritin and canine liver ferritin with H/L subunit ratios of 3.69 and 0.43, respectively. Canine IgM and IgA recognized both of the H- and L-subunit-rich isoferritins, showing that their binding activities to ferritin depend on the H-subunit content. Recombinant bovine H-chain ferritin homopolymer expressed in a baculovirus expression system bound more with IgM and IgA than the recombinant L-chain homopolymer expressed under the same conditions. These results suggest that canine IgM and IgA recognize H-subunit-rich isoferritins, and that H-subunit-rich isoferritins are cleared from the circulation more rapidly than L-subunit-rich isoferritins.     

Secreted ferritin subunits are of two kinds in insects molecular cloning of cDNAs encoding two major subunits of secreted ferritin from Calpodes ethlius

In insects, holoferritin is easily visible in the vacuolar system of tissues that filter the hemolymph and, at least in Lepidoptera, is abundant in the hemolymph. Sequences reported for insect secreted ferritins from Lepidoptera and Diptera have high sequence diversity. We examined the nature of this diversity for the first time by analyzing sequences of cDNAs encoding two ferritin subunits from one species, Calpodes ethlius (Lepidoptera, Hesperiidae). We found that insect secreted ferritin subunits are of two types with little resemblance to each other. Ferritin was isolated from iron loaded hemolymph of C. ethlius fifth instar larvae by differential centrifugation. The N-terminal amino acid sequences for the nonglycosylated subunit with Mr 24,000 (S) and the largest glycosylated subunit with Mr 31,000 (G) were determined. The N-termini of the two subunits were different and were used to construct degenerate PCR primers. The same cDNA products were amplified from cDNA libraries from the midgut which secretes holoferritin and from the fat body which secretes iron-poor apoferritin. The G subunit most closely resembles the glycosylated ferritin subunit from Manduca sexta and the S subunit resembles the Drosophila small subunit. The S and G subunits from Calpodes were dissimilar and distinct from the cytosolic ferritins of vertebrates and invertebrates. Additional sequences were obtained by 5' and 3' RACE from separate fat body and midgut RACE libraries. cDNAs encoding both subunits had a consensus iron responsive element (IRE) in a conserved cap-distal location of their 5' UTR. An integrin-binding RGD motif found in the G subunit and conserved in Manduca may facilitate iron uptake through a calreticulin (mobilferrin)/integrin pathway. Calpodes and other insect ferritins have conserved cysteine residues to which fatty acids can be linked. Dynamic acylation of ferritin may slow but not prevent its passage out of the ER.