mechanism of ferritin iron uptake: activity of the h-chain and

THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1988 by The American Society for Biochemistry and Molecular Biology, Inc

Vol. 263, No. 34, Issue of December 5, pp. 180&18092,1988 Printed in U. S. A.

Mechanism of Ferritin Iron Uptake: Activity of the H-chain and Deletion Mapping of the Ferro-oxidase Site A STUDY OF IRON UPTAKE AND FERRO-OXIDASE ACTIVITY OF HUMAN LIVER, RECOMBINANT H-CHAIN FERRITINS, AND OF TWO H-CHAIN DELETION MUTANTS*

(Received for publication, January 7, 1988)

Sonia LeviSQ, Alessandra Luzzagoll 11 **, Gianni Cesarenill, Anna CozziS, Francesca FranceschinelliS, Albert0 AlbertiniSS, and Paolo ArosioSOO From the $Department of Biomedical Sciences and Technology, University of Milano, Ospedale Sun Raffaele, Via Olgettina 60, I-20132 Milano, Italy, the YlEuropean Molecular Biology Laboratory, Meyerhofstrasse I , 0-6900 Heidelberg, Federal Republic of Germany, and the $$Institute of Chemistry, Faculty of Medicine, University of Brescia, Via Valsabbinu, I - 25100 Brescia, Italy

To study the functional differences between human ferritin H- and L-chains and the role of the protein shell in the formation and growth of the ferritin iron core, we have compared the kinetics of iron oxidation and uptake of ferritin purified from human liver (90% L) and of the H-chain homopolymer overproduced in Escherichia coli (100% H). As a control for iron autocatalytic activity, we analyzed the effect of Fe(II1) on the iron uptake reaction. The results show that the H- chain homopolymer has faster rates of iron uptake and iron oxidation than liver ferritin in all the conditions analyzed and that the difference is reduced in the conditions in which iron autocatalysis in high: i.e. at pH 7 and in presence of iron core. We have also analyzed the properties of two engineered H-chains, one lacking the last 22 amino acids at the carboxyl terminus and the other missing the first 13 residues at the amino terminus. These mutant proteins assemble in ferritin-like proteins and maintain the ability to catalyze iron oxidation. The deletion at the carboxyl terminus, however, prevents the formation of a stable iron core. It is concluded that the ferritin H-chain has an iron oxidation site which is separated from the sites of iron transfer and hydrolysis and that either the integrity of the molecule or the presence of the amino acid sequences forming the hydrophobic channel is necessary for iron core formation.

The major function of ferritin is to store and detoxify intracellular iron (1-3). Crystallographic studies have re- vealed that its structure is uniquely well designed to contain large amounts of iron in a soluble, nontoxic, and available form (1): it consists of a protein shell of 24 subunits arranged in 4-3-2 symmetry forming a cavity which is able to accom-

* The work done in Heidelberg was supported by European Com- munity Grant BAP 0252 (to G. C.). The work in Italy was partially supported by European Community Grant BAP 0246 (to P. A.) and by Regione Lombardia Grant 783. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accord- ance with 18 U.S.C. Section 1734 solely to indicate this fact.

5 Supported by a fellowship from the Instituto San Raffaele. 11 On leave of absence from Institute of Chemistry, University of

** Supported by an European Community training contract in the Brescia, Italy.

framework of the Biotechnology program. To whom correspondence should be addressed.

modate a core of up to 4500 atoms of hydrous ferric oxide polymers. The protein shell is pierced by eight channels lined by hydrophilic residues along the %fold symmetry axes and by six channels with hydrophobic properties along the 4-fold axes (1). Iron is believed to enter and exit the molecule through these channels.

The structural data depict this molecule as a sink for iron. However, the ferritin protein also plays an active role in promoting the formation and growth of the iron crystallite inside the cavity, since polynuclear iron complexes can form without the protein. I n vitro experiments have suggested that apoferritin requires ferrous iron to reconstitute ferritin core (4-6), and it is believed that this occurs in the cell as well.

Most of the present knowledge on apoferritin-iron interactions is derived from extensive studies on equine spleen ferritin. The role of the protein shell and the nature and order of the molecular events leading to iron core formation, however, are not completely understood. The in vitro reaction includes various chemico/physical events such as iron oxidation, transfer to the cavity (to the nucleation center or core surface), and hydrolysis/polymerization, but their order, kinetics, and localization in the molecule have not been clarified (4, 7-9). Earlier studies have shown that ferritin binds Fe(I1) as well as Fe(II1) in coordination to carboxyl residues (7, 10) and that iron nucleation occurs on carboxylate ligands that can be tentatively located in the cavity (11). Moreover, it was suggested that iron enters the molecule through the hydrophilic channels, by interactions with carboxyl ligands (1, 4).

Previous studies have shown that equine spleen ferritin is able to catalyze ferrous iron oxidation (ferro-oxidase activity) (12,13). This reaction requires molecular oxygen (8,14) which is not incorporated in the iron core (15) and is essential for ferritin core formation. However, it was recently shown that iron may enter equine spleen apoferritin in the reduced form and that oxidation follows as a slow process (9). Thus, it is presently unclear whether the protein shell plays a role in promoting iron oxidation (9) and, if so, whether it promotes the growth of the iron crystallite (16) or the crystallite itself drives its own growth (1, 17).

An interesting opportunity to study the mechanism of ferritin iron uptake is given by the presence in human cells of two distinct subunit types, named H (heavier) and L (lighter) that assemble in different proportions originating families of isoferritins (18). The amino acid sequences of H and L have been determined by sequencing the corresponding genes (19, 20) and show 55% homology, the most striking differences being present on the outer surface, in the cavity,

18086

Ferro-oxidase Activity of Ferritin H-chain 18087

and on the hydrophobic channel sequences (21, 22). By contrast, the hydrophilic channel appears to have identical sequences in the two chains (21). The analysis of the contribu- tion of the two chains to the functional differences observed in the various isoferritins should help understanding the mechanism of ferritin iron uptake.

We have recently cloned the cDNA for human ferritin H- chain into a plasmid vector that is able to direct the synthesis of large amounts of ferritin H-chain (rHF)’ that readily assembles in full H-24-mer ferritin (23). This “recombinant” ferritin has structural characteristics analogous to natural ferritin and is able to incorporate iron (23).

In the present work, we show that rHF can reach similar levels of iron saturation as natural human liver ferritin (HLF) that is composed by 90% L-chain. Moreover, rHF has faster rates of iron uptake than HLF in all the conditions studied and a higher ferro-oxidase activity in a reaction that uses ovotransferrin as final Fe(II1) acceptor (13). In this manu- script, we also describe the construction, purification, and preliminary characterization of two mutant H-ferritin homo- polymers. One, mutant rH-P161-am, that lacks the amino acid sequences that form the hydrophobic channel, assembles into a protein cage of molecular size similar to the wild type, and is able to oxidize iron, but it is unable to form a stable iron core. The second one, rH-D(Thrl-Hisl3), which contains a deletion of a flexible sequence exposed on the protein surface, assembles and incorporates iron similarly to wild- type ferritin. It is suggested that H-chain on the outer surface of the molecule carries a catalytic site that makes iron available to ovotransferrin and that activity of this site is inde- pendent from ferritin iron core formation and growth.

MATERIALS AND METHODS

Ferritins-Ferritin was purified from human liver as in Ref. 18. Recombinant H-ferritin was obtained by the Escherichia coli strain GC320 transformed with the plasmid pEMBLexSHFT, and ferritin expression was obtained as described previously (23) with the excep- tion that 1 mM iron nitrilotriacetate was added during induction t~ obtain iron-loaded recombinant ferritin, for easier purification. After growth and induction, the cells were precipitated, washed with 20 mM Tris, pH 7.4, 0.15 M NaCl (TBS), and stored at -20 “C or used immediately. The cells (about 10” in 20 ml) were disrupted by twice repeated sonication in TBS in an ice bath for 4 min (23). The ferritin was purified essentially as described previously (23): the homogenate was heated at 75 ‘C for 5 min, clarified, precipitated with ammonium sulfate (80% saturation), loaded on Sepharose 6B column, and precipitated by 90-min centrifugation at 200,000 X g. Typically from 1 liter of culture, 20-30 mg of purified ferritin were obtained, which showed a single band of 21 kDa in SDS electrophoresis (23).

Production and Purification of H-chain Mutants-To obtain the mutants rH-P161-am and rH-D(Thrl-Hisl3), the plasmid pEMB- Lex2HFT was subjected to site-directed mutagenesis (24) using oli- gonucleotides of sequence GATTACCCATGCAGGACTCAG and ATGGGAGCGTAGGAATCTGGC that introduce an amber muta- tion at position Pro-161 and a deletion of the sequence encoding the first 13 amino acids at the amino terminus, respectively. Mutant colonies were screened by hybridization to the mutagenic oligonucle- otides and the sequence of positive clones confirmed by the chain termination method (25).

E. coli strain GC320 transformed with the plasmids was grown and induced as described for wild type. Preliminary analysis using SDS electrophoresis showed that the expression of the mutants is at least as efficient as the wild type, accounting for about 15% of the total soluble bacterial proteins (not shown).

As with the wild type, the mutant rH-D(Thr1-Hisl3) is stable during the heating step at 75 “C and accumulates iron during expression in bacteria; thus, it was purified with success with the same

The abbreviations used are: rHF, recombinant H-chain ferritin; HLF, human liver ferritin; TBS, Tris-buffered saline; Hepes, 4-(2- hydroxyethy1)-1-piperazineethanesulfonic acid MES, 4-mOrphO- lineethanesulfonic acid.

procedure described above for rHF to obtain, with similar recovery, preparations judged pure by SDS electrophoresis.

The mutant rH-P161-am protein precipitates and denatures after heating at 75 “C and does not contain detectable iron, even when bacteria are grown in 1 mM iron nitrilotriacetate. Thus, for its purification the heat treatment and centrifugation at 200,000 X g steps cannot be applied, and a new purification procedure was devel- oped. During purification, the mutant protein was monitored by gel electrophoresis and by counter-immunoelectrophoresis with the anti- natural ferritin H-chain monoclonal antibody HF106 described previously (26). Growth, induction, and homogenization of the rH-P161- am clone were performed as for the wild type. The homogenate was added to ammonium sulfate at 80% saturation and the precipitate, after dialysis, was loaded on Sepharose 6B columns. The fractions containing the mutant protein were concentrated and loaded on Sephacryl S-200 columns equilibrated in TBS. For further purification, the sample was subjected to preparative gel electrophoresis as in Ref. 18 thus obtaining pure preparations, as judged by SDS electrophoresis.

Ferritins and their mutants were analyzed by electrophoresis in SDS or in nondenaturing polyacrylamide gels at pH 8.3 as described previously (18). The gels were stained with Coomassie Blue or with Prussian blue.

Preparation of Apoferritins-After purification and before iron uptake studies, iron was removed from HLF and rHF ferritin and the mutants as in Ref. 5, by incubation for 18 h in 1% thioglycolic acid, 0.1 M sodium acetate at pH 5.5 in a stoppered test tube. After disappearance of the color, excess 2,2’-bipyridine was added to chelate the ferrous iron and the sample extensively dialyzed against 0.1 M Hepes, pH 7. The apoferritins obtained in this way had a 280:260 absorbance ratio higher than 1.3.

Analytical Methods-Iron was determined as in Ref. 27: ferritin samples were incubated in 1 M acetic acid, 0.75 M sodium thiosulfite, 0.5% 2,2‘-bipyridine, boiled for 60 min, and the absorbance read at 520 nm. The molar extinction coefficient is 8030. Protein concentrations were determined with BCA reagent (Pierce Chemical Co.), using bovine serum albumin as standard. When the sample was in Hepes buffer, which interferes with the BCA assay, we used Bio-Rad protein assay (Bio-Rad).

Iron Core Formation-To aporHF, apoHLF, rH-P161-am, and bovine serum albumin (at a concentration of 0.05 mg/ml in 0.1 M Hepes buffer, pH 7.0) various amounts of ferrous ammonium sulfate were added and the solutions kept at 4 “C for 18 h. After the incubation, the solutions were centrifuged for 10 min at 10,000 X g and the supernatants analyzed for protein and iron concentration.

Iron Uptake-The kinetics of iron uptake were studied as in Ref. 29. Various amounts of freshly prepared ferrous ammonium sulfate dissolved in water at concentrations of 10-100 mM were added to various concentrations of apoferritins or of mutant proteins (0.1 p ~ , if not otherwise stated) in 0.1 M Hepes buffer or MES (for pH values below 6). The formation of amber iron products were followed in an Aminco dual wavelength spectrophotometer by monitoring the increase of optical density at 310 nm with reference at 600 nm to subtract the turbidity that develops when ferritin is absent. In experiments with Fe(III), freshly prepared solutions of ferric chloride or ferric nitrate in water were added to the protein-free Hepes buffer immediately before addition of ferrous ammonium sulfate. For cal- culation of the iron oxidized, we used the molar extinction coefficient of 2475 (28).

Ferro-oxidation-The ferro-oxidase activity of ferritin, using transferrin as Fe(II1) acceptor, was analyzed essentially as in Ref. 13. The ferritins and the mutants in the apo form, in 0.2 M sodium acetate, pH 6.0, with 4 mg/ml of apoovotransferrin (Recordati, Milano, Italy) were added to 0.1 mM ferrous ammonium sulfate, and the reaction followed either by differential spectra or by reading in the dual wavelength spectrophotometer at 470 nm with reference at 600 nm.

RESULTS

In order to gain information about the different contribu- tion of the H- and L-chain in the reaction that results in the formation of the iron core inside the ferritin cage, we have compared the properties of HLF (90% L) and rHF (100% H) by three different methods.

Size of the Iron Cores-To investigate how many iron atoms could be accumulated in the iron core of each ferritin molecule, rHF and HLF were incubated with various amounts of ferrous

18088 Ferro-oxidase Activity of Ferritin H-chain

iron for 18 h at 4 "C and the concentration of soluble iron and protein measured after centrifugation of the samples. In these conditions, all the non-ferritin polynuclear iron is quan- titatively precipitated, as can be deduced from control experiments where bovine serum albumin is added and ferritin is omitted. The results in Table I show that the two ferritins were able to reach similar saturation levels (3200-3300 iron atoms/molecule), rHF tends to precipitate at iron concentrations above 0.3 mM and HLF above 0.6 mM. The pH in the strongly buffered solution did not change after iron addition. Electrophoretic analysis of the samples after the incubation confirmed that all soluble iron comigrates with ferritin protein (not shown).

Iron Uptake Reactions-When ferritin is incubated in the presence of ferrous iron and dioxygen, an amber product is formed. The formation of this product can be monitored by measuring the absorbance of the reaction mix at 310 nm. We performed the reactions at low apoferritin (0.1 pM) and iron (0.1 mM) concentrations in order to have excess dioxygen (0.25 mM). The progression plots at pH 7.0 (Fig. 1) show faster rates of reaction in the presence of aporHF than in the presence of equal amounts of apoHLF. This latter reaction is only marginally faster than the formation of amber color in the absence of any added ferritin. Moreover, the reaction catalyzed by aporHF showed a hyperbolic progression plot, whereas in the presence of apoHLF, we observed a sigmoidal

TABLE I Iron incorporation in ferritin molecules

Apoferritin samples (50 Fg/ml) in 0.1 M Hepes, pH 7.0, were incubated with various amounts of ferrous ammonium sulfate for 18 h at 4 "C, centrifuged for 5 min at 10,000 X g to precipitate nonferritin polynuclear iron and the concentration of soluble iron and protein detected. In control experiments without protein or with 50 pg/ml bovine serum albumin, the soluble iron was always below 20 pM, as in the experiments with 50 pg/ml of the mutant rH-P161-am. ",: Iron ~ H F Saturation Iron HLF Saturation

PM PM PM tin PM PM ironjferri-

tin ironlferri-

0 0 0.10 0 0 0.11 0 100 117 0.10 1170 109 0.11 990 200 213 0.11 2123 187 0.11 1700 300 290 0.09 3222 283 0.11 2570 400 48 0.02 352 0.11 3200

1000 <20 co.01 (20 <0.01

shape as in the absence of ferritins. The three kinetics reached similar plateau levels, and after further addition of iron, the rates of amber color formation were faster and similar for the two ferritin samples.

TO be able to discern between the autocatalytic effect of the ferric iron formed during the incubation and the reaction specifically catalyzed by ferritin, we studied the rates of amber color formation at 310 nm in the presence of various amounts of ferric chloride or nitrate and without proteins. Fe(II1) catalyzes the reaction with progression plots similar to the ones obtained in the presence of ferritin, sigmoid when in excess of ferrous iron, and hyperbolic when in defect (Fig. 2). For further studies, we chose the Fe(II1) concentration of 50 pM that gave reaction rates similar to aporHF at pH 7. Cadmium and terbium inhibit ferritin iron uptake (12, 28, 30). Thus, we have explored the possibility to use these two inhibitors to distinguish between the autocatalytic and the ferritin-catalyzed reactions. We have observed that the amber color formation rates promoted by 50 pM Fe(III), 0.1 pM aporHF, and 0.1 p~ apoHLF are reduced to 11, 11, and 33%, respectively, when in the presence of 50 p M terbium and to 33,27, and 33%, respectively, when in the presence of 500 p M cadmium.

While analyzing apoferritin and iron-catalyzed amber color formation in the pH interval 5-7, we observed that the pH of the solution has a differential effect on the specific (apoferritin-catalyzed) and on the aspecific (Fe(II1)-catalyzed) reactions. At neutral pH, both reactions are comparable and fast in the conditions tested. At pH below 6.5, however, the reaction rate in the absence of ferritin decreases and becomes negligible below pH 6.0. By contrast, the two apoferritins, particularly the aporHF (Fig. 3), maintain significant reaction rates even at low pH. The results described until now would predict that the rate of iron uptake by ferritin should depend on the amount of ferric iron already present in the cage and that this effect should be pH-dependent. These predictions were proved correct when we studied the effect of loading with iron aporHF and apoHLF on the rates of iron uptake at two pH values (Fig. 4, A and 23). At pH 7, the presence of the iron core stimulates the uptake, reaching a maximum at 1500- 2000 iron atoms/molecule and with similar rates for both proteins, whereas at pH 6.5, the stimulatory effect of iron core is absent. At iron loading above 2000 iron atoms/molecule, an inhibitory effect was observed.

Ferritin Ferro-oxidase Enzymatic Actiuity-It was suggested that the ferro-oxidase activity of ferritin can be sepa-

" /

5

TIME cminl 10

FIG. 1. Progression plots of the iron uptake reactions in 0.1 M Hepes, pH 7.0, and 0.1 mM ferrous ammonium sulfate at 30 O C , the development of amber iron was followed at 310 nm. Dashed line is the control without protein. Line A: apoHLF, l ine B , apomutant rH-P161-am, line C, aporHF. After the solutions reached a plateau, a second aliquot of 0.1 mM Fe(I1) (arrow) was added.

0.2

E

0.1

!?

0

0 1 2

TIME Iminl

FIG. 2. Progression plots of amber color formation promoted by various amounts of ferric chloride (micromolar concentration shown in the figure) in absence of proteins, under the same conditions as Fig. 1.


01 0 I C .- E E \

C

s (3 p 0.01

0

5.0 5.1 6.0 e.1 7.0

pH FIG. 3. Initial rates of iron uptake at various pH values

with 0.1 mM ferrous ammonium sulfate and followed at 310 nm. Experiments were performed in 0.1 M Hepes at pH 6.5-7 and MES at pH 5-6. Samples: +, no protein; 0, 0.1 p M apoHLF; A, 50 p M FeCI3 and no protein; A, 0.1 p M aporH-D(Thr1-Hisl3); 0,O.l pM aporHF.

I 1 tobo 2000 3000

Fatw ATOMS/MOL. 1WO 2000 3000

Felw ATOMSIMOL.

FIG. 4. Initial velocities of iron uptake in 0.1 M Hepes with 0.1 mM ferrous ammonium sulfate of 0.1 p~ ferritin rHF (0) and HLF (0) preloaded from the apoproteins with various amounts of iron indicated as atoms of Fe(II1) per molecule of ferritin. A, reactions performed at pH 7.0. B, reactions at pH 6.5.

rated from the remaining steps leading to core formation by performing the reaction at pH 6.0 and including apotransferrin as an acceptor of the oxidized ferric iron (13). Formation of ferric iron can be monitored by measuring the Fe(II1). transferrin complex at 470 nm. Our studies were done in the conditions described previously (13), using apoovotransferrin (in place of apotransferrin) as the final acceptor of ferric iron. We confirm that, as previously observed (13), the reaction is of zero order with respect to transferrin in the concentration range 5-50 PM and is not affected by ferritin iron saturation (not shown). ApoHLF has an enzymatic activity slightly lower than equine spleen ferritin (13), in agreement with their structural similarities (18), whereas rHF activity is about 10- fold higher (Fig. 5 and Table 11).

Analysis of rH-PIGI-an and of rH-D(Thrl-HisI3) Mu- tants-The results described until now indicate that homo- polymers of H-ferritin have a higher ferro-oxidase activity than molecules rich in L-chain, suggesting that the catalytic site is not conserved in the two molecules. To further localize the region on the molecule that could be involved in the enzymatic activity, we have constructed two altered H-chains in which either the first 13 amino acids at the amino terminus or the last 22 amino acids, including the helix that forms the

T I M E lrn,",

FIG. 5. Progression plots of the formation of Fe(III).ovotransferrin complex monitored at 470 nm in the presence of 0.1 mM ferrous ammonium sulfate, 4 mg/ml apoovotransferrin, in 0.2 M sodium acetate pH 6.0. Dashed line is the control without ferritin. A, 0.2 p~ apoHLF B, 0.2 p~ apomutant rH-P161- am; C, 0.2 FM apomutant rH-D(Thr1-Hisl3); D, 0.2 p~ aporHF.

TABLE I1 Specific activities of ferritins and rH-mutants in ferro-oxidase assay

The formation Fe(II1)-ovotransferrin from 0.1 mM ferrous iron in presence of the apoproteins was followed at 470 nm for 3 min. The data are mean and standard deviations of quadruplicate experiments with aDoferritins and mutants at the concentration of 0.2 pM.

Irodferritin ~~

p M / m i n / p M

Apo-rHF 25.55 f 1.46

Apo-rH-P161-am 16.55 & 0.45 Apo-rH-D(Thr1-Hisl3) 17.49 f 0.47 Equine spleen ferritin" 5.7

APO-HLF 3.00 f 0.39

Deduced from Ref. 13, Fig. 2.

hydrophobic channel, have been deleted. Both these regions are not conserved in the two chains.

The two plasmids are able to express at high efficiency the peptides of the molecular mass of about 17 and 18 kDa for rH-D(Thr1-Hisl3) and rH-P161-am, respectively. From a preliminary characterization, it was evident that despite the extensive deletions, accounting for 6 and 12% of the total amino acid sequence, they still assemble in ferritin-like proteins; they migrate in nondenaturing gel electrophoresis, similar to ferritin, and elute from Sepharose 6B columns slightly after wild-type ferritin (V,/Vo of 1.65-1.69 versus 1.60 of rHF). In addition, both mutants are recognized by some monoclonal antibodies elicited by wild type, and free mutant peptides cannot be recognized in immunoblottings of the crude ho- mogenates.' The ferritin-like assembly is confirmed by the finding that the purified proteins in SDS electrophoresis show single peptides of 17-18 kDa (Fig. 6), whereas in nondenaturing electrophoresis they have a mobility similar to wild- type ferritin (Fig. 7). Despite a similar apparent assembly, the two mutants show distinctive characteristics: while the deletion of the first 13 amino acids does not affect the typical thermal stability and the capacity of the wild type to uptake iron during bacterial induction and can be purified with the same procedure, both characteristics are lost after deletion of

* S. Levi, A. Luzzago, G. Cesareni, A. Cozzi, F. Franceschinelli, A. Albertini, and P. Arosio, unpublished results.


1 2 3 -

-21 -1 9 -14

-I

FIG. 6. SDS electrophoresis in gradient pore acrylamide gels stained with Coomassie Blue of 20-pg samples. Lane 1, rH- P161-am; lane 2, rH-D(Thr1-Hisl3); lane 3, rHF. The positions of molecular weight standards are indicated.

1 2 3 4 5 6 7 8 -

+ A B

FIG. 7. Electrophoresis in 6% polyacrylamide gel run at pH 8.3. The purified apoferritin and apomutants samples were incubated for 30 min at room temperature with 500-fold molar excess of ferrous ammonium sulfate in 0.1 M Hepes, pH 7.0, run on the gel, and stained either with Coomassie Blue ( A ) or Prussian Blue (B) . Lanes 1 and 5, samples of 30 pg of HLF, stained with Coomassie and Prussian Blue, respectively; lanes 4 and 8,30 pg of rHF; lanes 3 and 7,30 pg of rH- D(Thr1-Hisl3); lanes 2 and 6, 30 pg of rH-P161-am. The arrow indicates the position of ferritin monomer.

the last 22 amino acids. This less stable protein can be purified with an alternative procedure.

The two purified mutants were subjected to iron kinetic studies. In the ferro-oxidase reaction, they showed similar and significant activity, accounting for about 70% of the wild type (Fig. 5 and Table 11). In the iron uptake system, the mutant rH-D(Thr1-Hisl3) has progression plots and activity analogous to the wild type in the pH range analyzed (Fig. 3). The mutant rH-pl61-am shows iron uptake kinetics, but at variance with rHF and HLF, further addition of iron after the reaction has reached plateau levels, does not increase the rate of iron uptake (Fig. 1).

In experiments designed to study iron loading, we found that rH-P161-am protein differs from wild type and rH-D(T1- H13) for not being able to incorporate and keep in solution iron, neither upon 18-h incubations (Table I) nor in shorter 30-min incubations (Fig. 7, lune 6).

DISCUSSION

In a previous report, we showed that the ferritin H-chain synthetized and assembled in E. coli has biochemical and immunological properties analogous to natural ferritin (23). Furthermore, this molecule is indistinguishable from natural ferritin H-chain in a series of assays such as inhibition of proliferation of progenitor myeloid cells (31) and binding to specific receptors on human cell lines (32). Here we show that this ferritin is able to incorporate ferrous iron, to keep ferric iron stably in solution, and to accumulate at saturation iron

levels similar to those of HLF. Thus, we conclude that it is a fully functional ferritin.

The comparison of the ferro-oxidase activity of the natural HLF, of the wild-type rHF, and two rHF mutants with extensive deletions at the amino or the carboxyl terminus allows us to exclude some region of the ferritin molecule that might contribute to the activity.

Iron Uptake and Oxidation-Ferritin iron uptake is usually analyzed by following the apoferritin-promoted formation of amber product from leuko ferrous iron (5, 6, 16, 17, 29). We performed the reactions in conditions in which, at equilib- rium, all the iron is incorporated into the ferritin in a soluble form, whereas in the absence of ferritin it is unstable and precipitates. Thus, the products obtained with and without the ferritins are different, but the progression plots are similar and reach the same plateau level (Fig. 1).

Ferritin acts as a true enzyme in the second reaction system used here, the product being the mononuclear pink complex Fe(II1). ovotransferrin (13). In this ferro-oxidation reaction, ferritin catalyzes the transformation of Fe(I1) into a form that is available to transferrin binding. This reaction follows events that occur on the outer surface of the molecule, and the significance of such events in the iron core formation has to be established. However, the parallel between this activity and iron uptake, which are both significantly faster for rHF and its mutants than for HLF, suggests that this reaction may be involved in the process of iron core formation.

The aporHF, composed by 100% H-chain, has rates of iron uptake and iron oxidation about 10-fold higher than apoHLF, composed by 10% H-chain. These figures are consistent with most of the ferro-oxidase activity being contributed by the H- subunit. AS a consequence, the assumption that equine spleen ferritin is, in a first approximation, equivalent to a homopolymer of the L-chain might not be valid, since the high activity of the low proportion of H-chain might obscure the contri- bution of the more abundant but less active L moiety.

The higher iron uptake activity of the H-chain described here is in agreement with previous analysis of natural human isoferritins with various H-subunit composition (33) and with the finding that the population of naturally occurring human isoferritins devoid of iron contain little or no H-chain (34, 35). This is suggestive of a different function, in cellular iron metabolism, of the H- and L-chains, in agreement with the hypothesis that the H-rich isoferritins may be primarily involved in iron detoxification (3) and intracellular iron trans- port (36).

Effect of Iron Autoxidation on Ferritin Iron Uptake-At pH 7, the presence of an iron core (up to 2000 molecules) has a strong stimulatory effect on iron uptake, and the differences between rHF and HLF are almost abolished. This is in agreement with the crystal growth model that stresses the active role played by the growing iron core (1). This effect, however, may be superimposed on the “true” ferritin catalytic properties and in some conditions may obscure it.

We have approached this problem by studying, in absence of proteins, the effect of Fe(II1) on amber color formation, the rationale being that Fe(II1) is formed during iron uptake and that mixtures of Fe(I1) and Fe(II1) are known to have strong chemical reactivity in lipid peroxidation (37).

Amber color formations promoted by Fe(II1) and by apoferritins have progression plots with sigmoid shape when ferrous iron is in large excess and a hyperbolic shape when it is in lower excess (Fig. 2 and Refs. 5 and 29), they both are inhibited by terbium and cadmium, and both are affected by pH in a similar way, decreasing dramatically in the pH interval 7.0-6.5 (Fig. 3). These findings pose some limits on


the interpretation of the ferritin “iron uptake” reaction; the sigmoid shape of the progression plot in the presence of ferritin was the first and major evidence for the crystal growth model, and the described effect of cadmium and terbium on ferritin (28, 30) may be primarily due to interaction of the inhibitory metals with iron and not with the protein shell.

The fast reaction rates of iron at pH 7.0 promoted by Fe(II1) in the absence of protein is consistent with the postulated stimulatory effect of iron core in ferritin iron uptake. At this pH, amber color formation is mainly driven by iron core, and the protein shell plays a passive role, in agreement with the “crystal growth” model (1). At pH 6.5, however, when the iron auto-oxidation rate is reduced, a preformed iron core has no stimulatory effect on iron uptake (Fig. 4B), indicating that at acidic pH, the crystal growth model does not apply (6, 16).

It is noteworthy that the difference in activity between apoHLF and aporHF is maximum in the initial part of the iron uptake reaction (;.e. when no Fe(II1) is present) and at pH 6.5, when iron autocatalysis is minimized. Thus, our data confirm the recently proposed model where ferritin iron uptake may switch from a crystal growth to a protein-mediated mechanism (4) and add that the conditions in which the switch occurs are related not only to Fe(I1) concentration as proposed (4), but to a competition between iron autocatalysis and a ferritin-promoted reaction. Since the rate of the former is reduced at low iron concentration, one may expect that in the cell where the free iron pool is exceedingly small, ferritin iron uptake is protein-mediated. This conclusion is further supported by previous studies that showed that the isoferritins richer in iron, that are present in a single tissue, are also richer in H-chain (38).

Mapping the Actiue Site-Our results indicate that ferritin H-chain has a ferro-oxidase catalytic activity. The ability of the two rHF mutants, in which either the first 13 or the last 22 amino acids have been deleted but that still assemble in stable ferritin-like proteins, to retain the catalytic activity indicates that the active site has not been deleted. Thus, the amino terminus or the hydrophobic channel do not play an essential role in the enzymatic reaction. An alternative localization of the active site in the hydrophilic channel can also be reasonably ruled out, since the sequences of H- and L- chains in this region are identical. As already suggested, the finding that transferrin can positively compete with the iron core for Fe(II1) during the reaction would reasonably exclude a localization of the active site in the interior of the cavity (13).

The involvement of carboxyl groups in ferritin iron uptake was first demonstrated by chemical modification studies (39) and confirmed with EPR (7) and extended x-ray absorption fine structure studies (11). Significant carboxyl residue sub- stitutions between H- and L-chains, conserved in all the species so far analyzed including chicken (40) and rat (41), are located on the loop and C helix on the outer surface Glu- 94 + Gly; Glu-101 --., Lys; Asp-123 + Ala) and in the inner cavity on the B and D helices (Lys-49 -+ Glu; His-57 + Glu; His-60 + Glu). The precise localization of the amino acid side chains that participate in the enzymatic reaction has to await the construction and characterization of new mutations directed to the amino acids that we have indicated.

The mutant rH-P161-am has the site for iron oxidation, but the sequences, which have been previously indicated as responsible for iron entry and nucleation (the hydrophilic channel and the inner cavity) ( l ) , have not been purposely modified. This protein, however, is not able to form a stable iron core. Although possible, it seems unlikely to us that the structure of the nucleation site in rh-P161-am is disrupted as

a consequence of a long distance effect of the deletion of 22 amino acids. In fact, the strong interactions along the 3- and 2-fold symmetry axes have not been modified, and the mutant assembles in a ferritin-like molecule. The simplest explana- tion of our results is that the amino acids in the E helix or the carboxyl-terminal tail are part of the nucleation site. An alternative hypothesis is that the deletion of the 4-fold hydrophobic channel opens large pores in the protein shell that drastically alter the chemical properties of the micro-environ- ment of the inner cavity. This mutant might prove to be a valuable tool to separate the enzymatic reaction (ferro-oxidase activity) from the subsequent steps that lead to ferritin iron core formation.

REFERENCES

1. Ford, G. C., Harrison, P. M., Rice, D. W., Smith, J. M. A., Treffry, A., White, J. L., and Yariv, J. (1984) Philos. Trans. R. SOC. Lond. B Biol. Sci. 304 , 551-565

2. Munro, H. N., and Linder, M. (1978) Physiol. Reu. 58,317-396 3. Drysdale, J. W. (1977) Ciba Found. Symp. 51,41-57 4. Harrison, P. M., Treffry, A., and Lilley, T. H. (1986) J. Znorg.

5. Macara, I. G., Hoy, T. G., and Harrison, P. M. (1972) Biochem.

6. Bryce, C. F. A., and Crichton, R. R. (1973) Biochem. J. 133,301-

7. Chasteen, N. D., and Theil, E. C. (1982) J. Biol. Chem. 2 5 7 ,

8. Goldner, B. C., Rinehart, A. L., Benshoff, H. M., and Harris, D. C. (1982) Biochim. Biophys. Acta 7 1 9 , 641-643

9. Rohrer, J. S., Joo, M.-S., Dartyge, E., Sayer, D. E., Fontaine, A., and Theil, E. C. (1987) J. Biol. Chem. 262,13385-13387

10. Wardeska, J. G., Viglione, B., and Chasteen, N. D. (1986) J. Biol.

11. Yang, C., Meagher, A., Huynh, B. H., Sayers, D. E., and Theil,

12. Niederer, W. (1970) Ezperientiu 2 6 , 218-220 13. Bakker, G. R., and Boyer, R. F. (1986) J. Biol. Chem. 261,13182-

14. Treffry, A., and Sowerby, J. M. (1979) FEBS Lett. 100,33-36 15. Meyer, D. E., Rohrer, J. S., Schoeller, D. A., and Harris, D. C .

16. Crichton, R. R., and Roman, F. (1978) J. Mol. Catal. 4, 75-82 17. Harrison, P. M., Hoy, T. G., Macara, T. G., and Hoare, J. (1974)

Biochem. J . 143, 445-451 18. Arosio, P., Adelman, T. G., and Drysdale, J. W. (1978) J. B i d .

Chem. 253,4451-4458 19. Costanzo, F., Colombo, M., Staempflis, S., Santoro, C., Marone,

M., Frank, R., Delius, H., and Cortese, R. (1986) Nucleic Acids Res. 14,721-736

20. Santoro, C., Marone, M., Ferrone, M., Costanzo, F., Colombo, M., Minganti, C., Cortese, R., and Silengo, L. (1986) Nucleic Acids Res. 14,2863-2876

21. Boyd, D., Vecoli, C., Belcher, D. M., Jain, S. K., and Drysdale, J. W. (1985) J. Biol. Chem. 260, 11755-11761

22. Leibold, E. A., Aziz, N. N., Brown, A. J. P., and Munro, H. N. (1984) J. Biol. Chem. 259,4327-4334

23. Levi, S., Cesareni, G., Arosio, P., Lorenzetti, R., Soria, M., Sollazzo, M., Albertini, A., and Cortese, R. (1987) Gene (Amst.)

24. Sollazzo, M., Frank, R., and Cesareni, G. (1985) Gene (Amst.)

25. Messing, J. (1983) Methods Enzymol. 101, 20-78 26. Luzzago, A., Arosio, P., Iacobello, C., Ruggeri, G., Capucci, L.,

Brocchi, E., De Simone, F., Gamba, D., Gabri, E., Levi, S., and Albertini, A. (1986) Biochim. Biophys. Acta 8 7 2 , 61-71

27. Niitsu, I., and Listowsky, 1. (1973) Biochemistry 12 , 4690-4695 28. Macara, I. G., Hoy, T. G., and Harrison, P. M. (1973) Biochem.

29. Paques, E. P., Paques, A., and Crichton, R. R. (1980) Eur. J.

30. Treffry, A., and Harrison, P. M. (1984) J. Znorg. Biochem. 2 1 , 9 -

31. Broxmeyer, H. E., Lu, L., Bicknell, D. C., Williams, D. E., Cooper,

Biochem. 27,287-293

J. 126,151-162

309

7672-7677

Chem. 26 1,6677-6683

E. C. (1987) Biochemistry 26 , 497-503

13185

(1983) Biochemistry 22,876-880

5 1, 267-272

37,199-206

J. 135, 785-789

Biochem. 107,447-453

20


S., Levi, S., Salfeld, J., and Arosio, P. (1986) Blood 6 8 , 1257- 1263

Ciriello, M. M., Arosio, P., and Barosi, G. (1983) Blood 6 2 , 1078-1087

32. Fargion, S., Arosio, P., Fracanzani, A. L., Cislaghi, V., Levi, S., 37. Braughler, J. M., Duncan, L. A., and Chase, R. L. (1986) J . Bzol. Cozzi, A., Piperno, A., and Fiorelli, G. (1988) Blood 71 , 753- 757

Chem. 261,10282-10289 38. Ishitani, K., Listowsky, I., Hazard, J., and Drysdale, J. W. (1975)

173,969-977 39. Wetz, K., and Crichton, R. R. (1976) Eur. J. Biochem. 6 1 , 545- 33. Wagstaff, M., Wonvood, M., and Jacobs, A. (1978) Biochem. J. J. Biol. Chem. 2 5 0 , 5446-5449

34. Arosio, P., Yokota, M., and Drysdale, J. W. (1977) Br. J. Hue- 550

35. Santambrogio, P., Cozzi, A,, Levi, S., and Arosio, P. (1987) Br. J. Bid . 7, 1751-1758

36. Cazzola, M., Dezza, L., Bergamaschi, G., Belotti, V., Caldera, D., Acad. Sci. U. S. A. 8 4 , 7438-7442

matol. 36 , 199-207 40. Stevens, P. W., Dodgson, J. B., and Engel, J. D. (1987) Mol. Cell.

Haematol. 6 5 , 235-237 41. Murray M. T., White, K., and Munro, H. N. (1987) Proc. Natl.

mechanism of ferritin iron uptake: activity of the h-chain and

Documents