Iron in phytoferritin from legume seeds is required for seedling germination and early growth. However, the mechanism by which phytoferritin regulates its iron complement to these physiological processes remains unknown. In the present study, protein degradation is found to occur in purified SSF (soya bean seed ferritin) (consisting of H-1 and H-2 subunits) during storage, consistent with previous results that such degradation also occurs during seedling germination. In contrast, no degradation is observed with animal ferritin under identical conditions, suggesting that SSF autodegradation might be due to the EP (extension peptide) on the exterior surface of the protein, a specific domain found only in phytoferritin. Indeed, EP-deleted SSF becomes stable, confirming the above hypothesis. Further support comes from a protease activity assay showing that EP-1 (corresponding to the EP of the H-1 subunit) exhibits significant serine protease-like activity, whereas the activity of EP-2 (corresponding to the EP of the H-2 subunit) is much weaker. Consistent with the observation above, rH-1 (recombinant H-1 ferritin) is prone to degradation, whereas its analogue, rH-2, becomes very stable under identical conditions. This demonstrates that SSF degradation mainly originates from the serine protease-like activity of EP-1. Associated with EP degradation is a considerable increase in the rate of iron release from SSF induced by ascorbate in the amyloplast (pH range, 5.8–6.1). Thus phytoferritin may have facilitated the evolution of the specific domain to control its iron complement in response to cell iron need in the seedling stage.

INTRODUCTION

Ferritins are a class of multimeric iron storage and detoxification proteins. The importance of their dual function is underscored by their ubiquitous distribution throughout all organisms, with the exception of fungi. The ferritin complex has 24 subunits assembled into a spherical shell characterized by a 432-point symmetry. Up to 4500 Fe3+ atoms can be stored either as the crystalline mineral ferrihydrite or as amorphous hydrous ferric oxyphosphate in the inner cavity of the assembled ferritin shell [13]. Structural analyses of vertebrate, plant and bacterial ferritins indicate that each subunit consists of a four-helix bundle (helices A, B, C and D) and a fifth short helix (helix E) [1,4,5]. In mammals, two distinct ferritin subunits (H and L) are found with similar three-dimensional structures. The H subunit has ferroxidase centres responsible for fast Fe2+oxidation. In contrast, L subunits lack ferroxidase centres, and thus do not exhibit fast Fe2+ oxidation kinetics, but facilitate nucleation of the mineral core [5].

Amino acids involved in the definition of the ferroxidase centre are strictly conserved in all plant ferritins except for PSF (pea seed ferritin), where a histidine residue is replaced with a glutamic acid residue at position 62 of the amino acid sequence in the ferroxidase centre [6,7]. However, plant and animal ferritins are remarkably different in their cytological localization. In contrast with animal ferritin existing in the cell cytoplasm, plants store iron within ferritin mainly in plastids, such as the amyloplast in seeds [3]. Unlike animal ferritin, in which two types of subunits (H and L) occur, only the H-type subunit has been described in phytoferritin. Ferritin from dried soya bean seed consists of two subunits, a 26.5 kDa (H-1) subunit and a 28.0 kDa (H-2) subunit, which share ~80% amino acid sequence identity [8,9]. The two subunits are encoded by two distinct genes, SferH-1 (GenBank® accession number M64337) and SferH-2 (GenBank® accession number AB062754) respectively [8], and are synthesized as a precursor (32 kDa) that contains a TP (transit peptide) and a following EP (extension peptide) at its N-terminal. The TP is responsible for the precursor targeting plastids [10]. Upon transport to the plastids, the TP is cleaved from the subunit precursor, producing the mature subunit which assembles in a 24-mer ferritin within the plastids [11]. Thus in mature phytoferritin, 24 EP domains per molecule represent the major structural difference between animal and plant ferritin. In PSF, each EP domain is composed of 24 amino acid residues, 11 of which form a specific α-helix, termed the P-helix, flanked by proline residues (X and L) [6]. Differing from the TP in function, it was recently discovered that the EP serves as a second binding and ferroxidase centre contributing to iron core mineralization at high ferritin iron loadings (>48 iron/protein shell) [12], indicative of the role of the EP in iron oxidative deposition in phytoferritin. Although Arabidopsis ferritin is considered crucial to protecting cells against oxidative damage rather than for iron storage [13], in legume seeds, the majority of total iron is stored in ferritin in the amyloplast; such storage is known to meet plant demand during seedling germination and growth [1,14]. Therefore elucidating whether or not the EP plays a role in phytoferritin iron release in the seedling stage is the focus of the present study.

In the present study, protein degradation was found to occur upon SSF (soya bean seed ferritin) standing at 4 °C, an observation consistent with a previous report where PSF showed the same protein degradation taking place during seed germination [7,14]. Further studies revealed that the EP of SSF has serine protease-like activity, which causes its degradation, probably because of its exposure to the exterior surface of the protein. Interestingly, removal of the EP from WT (wild-type) SSF considerably facilitates iron release from ferritin with the presence of ascorbate in the pH range 5.8–6.1, demonstrating that the EP is closely associated with iron release from phytoferritin.

EXPERIMENTAL

Reagents

All chemicals used were of reagent grade or purer. Boc-Gln-Ala-Arg-MCA [where Boc is t-butoxycarbonyl and MCA is (7-methoxycoumarin-4-yl)acetyl] and N-succinyl-Ala-Phe-Lys-MCA were purchased from Sigma–Aldrich. PMSF, AEBSF [4-(2-aminoethyl)benzenesulfonyl fluoride], antipain, EDTA, pepstatin, benzamidine and leupeptin were from Amresco.

Preparation of WT SSF, recombinant SSF and EP-deleted SSF

SSF was purified [12,15] and rH-1 (recombinant H-1) prepared [8,16] as described previously. The expression plasmid for rH-2 (recombinant H-2) was constructed by inserting the H-2 cDNA into the NcoI/BamHI site of pET21d using a PCR-based method. Constructs were then introduced into the Escherichia coli strain BL21 (DE3). Positive transformants of each construct were grown at 37 °C on an LB (Luria–Bertani) medium supplemented with 50 mg/l carbenicillin, and protein expression was induced with 100 μM IPTG (isopropyl β-D-thiogalactoside) when cell density reached an A600 of 0.6. Both rH-1 and rH-2 were purified using the same methods described above for native SSF. Apoferritin was prepared as described previously [17,18]. The concentrations of all ferritin types were determined according to the Lowry method with BSA as a standard. SDS/PAGE was performed under reducing conditions using 15% mini-slab gels according to a previously reported method [19]. Preparation and identification of SSF with the EP deleted was carried out as reported recently [12].

CD and fluorescence spectroscopy

CD spectra were recorded on a PiStar-180 spectrometer (Applied Photophysics) at 25 °C under a constant flow of nitrogen gas. Typically, a cell with a 0.1-cm pathlength was used for spectral measurements between 190 and 260 nm. The spectra represent an average of 6–10 scans. CD intensities reported in the figures are expressed in Δε (M−1·cm−1). Fluorescence spectra were measured using a Cary Eclipse spectrofluorimeter (Varian) at 25 °C with 280 nm as an excitation wavelength. The spectral resolution was 1.0 nm.

Effect of protease inhibitors on SSF degradation

Different kinds of protease inhibitors were added to the SSF solution to obtain the final concentration (5 mM PMSF, 5 mM EDTA, 20 μM pepstatin, 2 mM benzamidine and 20 μM leupeptin). The effective inhibiting concentrations of the protease inhibitors were used as described previously [20,21]. After incubation of the mixture at 4 °C for 40–50 days, samples were applied to SDS/PAGE. A control test was performed without the addition of inhibitors.

Determination of EP protease-like activity

Potential serine protease activity was determined as previously described [21,22] with slight modifications. Enzyme assays using peptide–MCA substrates (Boc-Gln-Ala-Arg-MCA and N-succinyl-Ala-Phe-Lys-MCA) were performed by fluorometric determination of liberated AMC (7-amino-4-methylcoumarin). Briefly, 3.94 ml of 50 mM Tris/HCl buffer solution (pH 8.0) containing 100 mM NaCl and 10 mM CaCl2 was added to 40 μl of peptide MCA substrate dissolved in DMSO (10 mM), followed by mixing with 20 μl of EP into a fluorescence cuvette at 25 °C. The AMC liberation by enzymatic hydrolysis was monitored with a Cary Eclipse spectrofluorimeter (Varian) at 25 °C for 120 s. Fluorescence was measured using an excitation wavelength of 380 nm and an emission wavelength of 460 nm [22]. Control samples were created under the same conditions, except that the above protein solution was replaced with either 20 μl of BSA (40 μg) or 20 μl of Alcalase (1000-fold dilution of Alcalase 2.4L).

Reactions between the EP and PMSF

A 1 μl aliquot of 140 mM PMSF was added to 100 μl of EP-1 (140 μM) in 5 mM Mops at pH 7.0, and the resulting solution was incubated overnight at 4 °C. Subsequently, the solution was mixed with 0.1% TFA (trifluoroacetic acid) and 10 mg/ml CHCA (α-cyano-4-hydroxy cinnamic acid) prior to MALDI–TOF-MS (matrix-assisted laser-desorption ionization–time-of-flight MS) analyses.

The degraded products of the ~24.5 and 26.5 kDa SSF subunits were separated electrophoretically using SDS/PAGE (15% gels). Gel spots were prepared as described previously [9]. All mass spectra of MALDI–TOF-MS were obtained on a Bruker Ultraflex III TOF/TOF (Bruker Daltonik) in positive-ion mode at an accelerating voltage of 20 kV with a nitrogen laser (337 nm) [12].

Kinetics of iron release from holoferritin

Iron release from SSF was investigated with a stopped-flow instrument (a Hi-Tech SFA-20M apparatus) in conjunction with a Cary 50 spectrophotometer (Varian) using the assay procedure described previously [23]. All concentrations stated were final after mixing the two reagents. The mixing dead time was determined to be 6.8±0.5 ms using the DICP (dichloroindophenol) and ascorbic acid test reaction [24]. The development of [Fe(ferrozine)3]2+ was measured by recording the increase in absorbance at 562 nm, and the iron released estimated using ε562=27900 M−1·cm−1 [25]. The initial rate (ν0) of iron release was measured as described previously [26].

Preparation and pH measurement of amyloplasts from soya bean seed cotyledons

Amyloplasts were prepared according to previously reported methods, with slight modifications [27,28]. Soya bean seeds were imbibed in ddH2O (double-distilled H2O) at room temperature (25 °C) for 12 h in the dark and sown in a petri dish. After germination at 12, 48 and 72 h respectively, amyloplasts were isolated from cotyledons, which were chopped with a razor blade and immediately frozen with liquid nitrogen. The cotyledon slices were then homogenized with 10-fold ultrapure water in a Waring Blendor. This homogenate was quickly filtered through a double layer of nylon mesh (300 mesh). The filtrate was centrifuged at 500 g for 10 min at 4 °C. The pellets were washed 6–8 times with ultrapure water and were centrifuged under the same conditions as above until the intact amyloplast was examined by light microscopy (Supplementary Figure S6B). The amyloplast pellets were crushed using an oscillator for 10 min. After centrifugation at 13200 g for 20 min at 4 °C, the pH value of the supernatant was determined with a microelectrode (MI 402; Microelectrodes).

RESULTS

Autodegradation of SSF

SSF was purified to homogeneity with an apparent molecular mass estimated to be approx. 560 kDa by native PAGE (Supplementary Figure S1A at http://www.BiochemJ.org/bj/427/bj4270313add.htm). Furthermore, SDS/PAGE analysis indicated that the ferrritin complex contained two kinds of subunits (28.0 and 26.5 kDa) (Supplementary Figure S1B) present in purified native ferritin in an approximately 1:1 ratio, as reported previously [8]. To determine whether SSF was stable, the protein was allowed to stand at 4 °C for 10–50 days. Since SSF exhibits good solubility at pH 8.0, 50 mM phosphate buffer containing 150 mM NaCl (buffer B) at pH 8.0 was used as a sample buffer. As shown in Figure 1, two ferritin subunits began to degrade into ~24.5 kDa polypeptide(s) after 10 days of incubation. As the incubation time increased from 10 to 50 days, degradation significantly increased. Similar results were obtained with other buffers, such as Mops or Mes in the 6.0–7.5 pH range (results not shown), and protein samples from three different protein preparations/purifications, suggesting that the observed degradation is not sample/buffer/pH-dependent. In contrast, such protein degradation was not observed in animal ferritins such as HoSF (horse spleen ferritin) and HuHF (human H-chain ferritin) under the same experimental conditions (results not shown).

SDS/PAGE analysis of SSF degradation

Figure 1
SDS/PAGE analysis of SSF degradation

The purified SSF was incubated in buffer B at 4 °C at different time intervals. Lanes 0–5 correspond to incubation times of 0, 10, 20, 30, 40 and 50 days respectively. Lane M, protein markers and their corresponding molecular masses.

Figure 1
SDS/PAGE analysis of SSF degradation

The purified SSF was incubated in buffer B at 4 °C at different time intervals. Lanes 0–5 correspond to incubation times of 0, 10, 20, 30, 40 and 50 days respectively. Lane M, protein markers and their corresponding molecular masses.

No autodegradation with EP-deleted SSF

The EP represents the major structural difference between plant and animal ferritins [1,8,12], raising a question as to whether the EP is involved in SSF autodegradation. To answer this question, EP-deleted SSF was prepared by incubating WT SSF with a commercially available protease, Alcalase 2.4L, at 60 °C for 5 min. The newly prepared holoSSF has a molecular mass of ~440 kDa (Figure 2A, lane 1) and contains the same amount of iron within the interior core as WT holoSSF (results not shown). SDS/PAGE analysis indicated that the 28.0 and 26.5 kDa subunits were completely hydrolysed to two small subunits with molecular masses of 23.5 and 21.0 kDa respectively (Figure 2B, lane 1). Subsequently, the first 15 amino acid residues of the N-terminal of the two new subunits were determined (and are listed in Figure 3B), further confirming removal of the EP. Moreover, both WT and EP-deleted SSF had nearly the same CD (Supplementary Figure S2A at http://www.BiochemJ.org/bj/427/bj4270313add.htm) and fluorescence spectra (Supplementary Figure S2B), indicating that the structure of EP-deleted SSF is similar to that of WT SSF.

Native PAGE and SDS/PAGE analyses of SSF with the EP deleted

Figure 2
Native PAGE and SDS/PAGE analyses of SSF with the EP deleted

(A) Native PAGE at pH 8.3. (B) SDS/PAGE at pH 8.3. Lane 1, SSF with the EP deleted (molecular mass 440 kDa); lane 2, wild-type SSF (molecular mass 560 kDa); and lane M, protein markers and their corresponding molecular masses.

Figure 2
Native PAGE and SDS/PAGE analyses of SSF with the EP deleted

(A) Native PAGE at pH 8.3. (B) SDS/PAGE at pH 8.3. Lane 1, SSF with the EP deleted (molecular mass 440 kDa); lane 2, wild-type SSF (molecular mass 560 kDa); and lane M, protein markers and their corresponding molecular masses.

Amino acid sequence of (A) WT SSF and (B) 15 N-terminal sequence residues of two subunits of SSF whose EP has been deleted by Alcalase 2.4L

Figure 3
Amino acid sequence of (A) WT SSF and (B) 15 N-terminal sequence residues of two subunits of SSF whose EP has been deleted by Alcalase 2.4L

The EP is indicated in boldface.

Figure 3
Amino acid sequence of (A) WT SSF and (B) 15 N-terminal sequence residues of two subunits of SSF whose EP has been deleted by Alcalase 2.4L

The EP is indicated in boldface.

To compare the stability of the two proteins, experiments were conducted wherein both WT and EP-deleted SSF were allowed to stand at 4 °C over a period of 60 days. As observed above, protein degradation occurred pronouncedly in both WT holoSSF and apoSSF (Figure 4, lanes 1 and 2), whereas EP-deleted holoSSF and apoSSF were relatively stable (Figure 4, lanes 3 and 4). This points to a functional role for the EP that is associated with protein degradation through a certain way (see below). Even upon standing for 60 days or longer, no subunits with a molecular mass less than 23.5 and 21.0 kDa were generated (results not shown), indicating that the EP itself is susceptible to degradation. In addition, WT apoSSF was also unstable, indicating that iron was not involved in such degradation, or at least was not a main factor.

Comparison of the stability of WT SSF and SSF with the EP deleted by SDS/PAGE analysis

Figure 4
Comparison of the stability of WT SSF and SSF with the EP deleted by SDS/PAGE analysis

The samples were incubated in buffer B at 4 °C for 60 days. Lane 1, holoSSF; lane 2, apoSSF; lane 3, holoSSF with the EP deleted; lane 4, apoSSF with the EP deleted; and lane M, protein markers and their corresponding molecular masses.

Figure 4
Comparison of the stability of WT SSF and SSF with the EP deleted by SDS/PAGE analysis

The samples were incubated in buffer B at 4 °C for 60 days. Lane 1, holoSSF; lane 2, apoSSF; lane 3, holoSSF with the EP deleted; lane 4, apoSSF with the EP deleted; and lane M, protein markers and their corresponding molecular masses.

Effect of different kinds of protease inhibitors on SSF degradation

The above observation raises the possibility that the EP has protease-like activity. To confirm this, SSF was mixed with different kinds of protease inhibitors followed by standing at 4 °C for 40 days. Final concentrations of these inhibitors were used as described previously [20,21]. As shown in Figure 5(A), holoSSF degradation was nearly completely inhibited by a serine protease inhibitor, PMSF, whereas EDTA (a metalloprotease inhibitor), pepstatin (an aspartic protease inhibitor) and leupeptin (a serine protease and cysteine protease inhibitor) hardly had any inhibitory effect on the degradation. A trypsin-like serine protease inhibitor, benzamidine, did not display an inhibitory effect. Moreover, the combination of PMSF with EDTA or other inhibitors did not exhibit an enhanced inhibitory effect. As expected, other serine protease inhibitors, such as AEBSF and antipain, also exhibited nearly identical effects with PMSF (Supplementary Figure S3 at http://www.BiochemJ.org/bj/427/bj4270313add.htm). Thus it can be concluded from these results that the activity of serine proteases is responsible for the observed degradation of the phytoferritin EP. Nearly the same results were obtained after incubation of these protease inhibitors with SSF for 50 days at 4 °C (Figure 5B), further supporting this conclusion.

Effect of different kinds of protease inhibitors on SSF degradation

Figure 5
Effect of different kinds of protease inhibitors on SSF degradation

(A) The purified SSF was incubated in buffer B at 4 °C for 40 days; (B) the sample was incubated in buffer B at 4 °C for 50 days. Lane 1, 5 mM EDTA; lane 2, 5 mM PMSF; lane 3, 20 μM pepstatin; lane 4, 2 mM benzamidine; lane 5, 20 μM leupeptin; lane 6, 5 mM EDTA and 5 mM PMSF; lane 7, 5 mM EDTA and 20 μM pepstatin; lane 8, 5 mM EDTA, 5 mM PMSF and 20 μM pepstatin; lane 9, control (without protease inhibitors); and lane M, protein markers and their corresponding molecular masses.

Figure 5
Effect of different kinds of protease inhibitors on SSF degradation

(A) The purified SSF was incubated in buffer B at 4 °C for 40 days; (B) the sample was incubated in buffer B at 4 °C for 50 days. Lane 1, 5 mM EDTA; lane 2, 5 mM PMSF; lane 3, 20 μM pepstatin; lane 4, 2 mM benzamidine; lane 5, 20 μM leupeptin; lane 6, 5 mM EDTA and 5 mM PMSF; lane 7, 5 mM EDTA and 20 μM pepstatin; lane 8, 5 mM EDTA, 5 mM PMSF and 20 μM pepstatin; lane 9, control (without protease inhibitors); and lane M, protein markers and their corresponding molecular masses.

To determine what concentration of PMSF is effective to prevent SSF degradation, a series of PMSF solutions with different concentrations were tested. As shown in Supplementary Figure S4 (at http://www.BiochemJ.org/bj/427/bj4270313add.htm), SSF degradation was partially inactivated by 0.5 mM PMSF. In contrast, when the PMSF concentration reached 1 mM, a nearly complete inhibition was obtained. Therefore 1 mM PMSF was used during the isolation and purification of SSF.

Serine protease-like activity of the EP domain of SSF

To seek direct evidence of where the serine protease-like activity originates from, both EP-1 with the sequence ASTVPLTGVIFEPFEEVKKSELAVPT (corresponding to the EP from the H-1 subunit) and EP-2 with the sequence ASNAPAPLAGVIFEPFQELKKDYLAVPI (corresponding to the EP from the H-2 subunit) were synthesized and their hydrolysing activity was measured with Alcalase (1000-fold dilution of Alcalase 2.4L) and BSA as control samples. As expected, BSA has almost no activity against the two peptides, whereas Alcalase had the strongest activity among all tested samples (Figure 6). Similar to BSA, EP-2 had a much weaker activity compared with the analogue EP-1, which showed good activity against Boc-Gln-Ala-Arg-MCA (Figure 6A) and N-succinyl-Ala-Phe-Lys-MCA (Figure 6B), indicating that EP-1 is mainly responsible for the EP hydrolysing activity. Furthermore, it was observed that the combination of EP-1 and EP-2 exhibited marginally stronger activity than EP-1 alone. Similar to other serine proteases, EP-1 also revealed a stronger activity against Boc-Gln-Ala-Arg-MCA than against N-succinyl-Ala-Phe-Lys-MCA [21]. Although the relative catalytic activity of EP-1 is profoundly stronger than that of EP-2, its absolute catalytic activity is still very low, excluding the characterization of enzymology for EP-1.

Relative enzyme activities of EP-2, EP-1, and EP-2/EP-1 (1:1) towards two peptide–MCA substrates, Boc-Gln-Ala-Arg-MCA (A) and N-succinyl-Ala-Phe-Lys-MCA (B)

Figure 6
Relative enzyme activities of EP-2, EP-1, and EP-2/EP-1 (1:1) towards two peptide–MCA substrates, Boc-Gln-Ala-Arg-MCA (A) and N-succinyl-Ala-Phe-Lys-MCA (B)

BSA and Alcalase (1000-fold dilution of Alcalase 2.4L) were used as control samples. Conditions: 0.7 μM EP-2, EP-1 or EP-2/EP-1 (1:1), 0.1 mM Boc-Gln-Ala-Arg-MCA or N-succinyl-Ala-Phe-Lys-MCA, in 50 mM Tris/HCl solution (pH 8.0) containing 100 mM NaCl and 10 mM CaCl2. Values are expressed as means±S.D. (n=3).

Figure 6
Relative enzyme activities of EP-2, EP-1, and EP-2/EP-1 (1:1) towards two peptide–MCA substrates, Boc-Gln-Ala-Arg-MCA (A) and N-succinyl-Ala-Phe-Lys-MCA (B)

BSA and Alcalase (1000-fold dilution of Alcalase 2.4L) were used as control samples. Conditions: 0.7 μM EP-2, EP-1 or EP-2/EP-1 (1:1), 0.1 mM Boc-Gln-Ala-Arg-MCA or N-succinyl-Ala-Phe-Lys-MCA, in 50 mM Tris/HCl solution (pH 8.0) containing 100 mM NaCl and 10 mM CaCl2. Values are expressed as means±S.D. (n=3).

To gain insight into the mechanism by which PMSF prevents SSF degradation, the interaction of EP-1 with PMSF was studied by MALDI–TOF-MS. The MS profile of the intact EP-1 showed a single peak at m/z 2790.120 Da (Figure 7A). This indicates a singly charged peptide monomer [M+H]+ of the peptide, which is in good agreement with the predicted molecular mass of 2790.240 Da. This peak rose to 2810.474 Da upon treatment with PMSF (Figure 7B). The difference in molecular mass between treated EP and untreated EP is 20.354 Da, which is approximately equal to the mass of one molecule of hydrogen fluoride (20.006 Da), which is derived from PMSF hydrolysis [29]. Therefore the observed inhibition of EP degradation by PMSF might be derived from the combination of EP-1 and hydrogen fluoride. This result provided a good explanation for why PMSF can completely inhibit SSF degradation.

MS profiles of EP-1 (A) and the EP treated with PMSF (PMSF/EP=10:1) (B) acquired by MALDI–TOF-MS

Comparison of stability of rH-1 and rH-2 SSF

To seek further evidence of the protease activity of EP-1, the stabilities of rH-2 and rH-1 were compared by SDS/PAGE analysis. Similar to WT SSF, rH-1 also began to degrade into smaller polypeptides after 20 days of incubation at 25 °C (Figure 8B), whereas rH-2 was very stable (Figure 8A), with no degradation under the same conditions. This confirms the above conclusion that EP-1, but not EP-2, has the serine protease-like activity responsible for the observed SSF autodegradation during storage. Consistent with the present observations, MALDI–TOF-MS results showed that the peptide mass fingerprint of the degraded product of SSF with a molecular mass of ~24.5 kDa (Figure 1) corresponded significantly with that of the 26.5 kDa SSF subunit (Supplementary Figure S5 at http://www.BiochemJ.org/bj/427/bj4270313add.htm), as suggested by its high probability-based Mowse score (131). This, once again, demonstrates that the H-1 subunit is prone to degradation.

Comparison of the stability of WT SSF, rH-2 and rH-1 by SDS/PAGE analysis

Figure 8
Comparison of the stability of WT SSF, rH-2 and rH-1 by SDS/PAGE analysis

(A) The samples were incubated in buffer B at 25 °C for 20 days. Lane 1, WT SSF; lane 2, rH-2; and lane M, protein markers and their corresponding molecular masses. (B) Lane 1, rH-1 incubated in buffer B at 25 °C for 1 day; lane 2, rH-1 incubated in buffer B at 25 °C for 20 days; and lane M, protein markers and their corresponding molecular masses.

Figure 8
Comparison of the stability of WT SSF, rH-2 and rH-1 by SDS/PAGE analysis

(A) The samples were incubated in buffer B at 25 °C for 20 days. Lane 1, WT SSF; lane 2, rH-2; and lane M, protein markers and their corresponding molecular masses. (B) Lane 1, rH-1 incubated in buffer B at 25 °C for 1 day; lane 2, rH-1 incubated in buffer B at 25 °C for 20 days; and lane M, protein markers and their corresponding molecular masses.

Comparison of the rate of iron release from WT and EP-deleted holoSSF

To shed light on the physiological function of EP autodegradation, the rate of iron release from WT holoSSF induced by ascorbic acid was compared with that from EP-deleted holoSSF. To determine what pH values should be used for the iron release from SSF, the pH of amyloplasts isolated from SSF during seedling germination and growth was directly measured as described previously [27,28]. It was found that the amyloplast pH was in the range 5.8–6.1 (Supplementary Figure S6 at http://www.BiochemJ.org/bj/427/bj4270313add.htm) at different germination periods. Therefore the level of ferrous atoms released from the ferritin shell was monitored at pH≤6.5 by the formation of the Fe2+–ferrozine complex [23]. Generally, iron release from EP-deleted holoSSF was much faster than that from WT holoSSF at all pH values tested (Table 1). For example, at pH 6.0, the initial rate of iron release from EP-deleted holoSSF (2.36±0.15 μM/s) is 12-fold larger than that from EP-deleted holoSSF (0.194±0.013 μM/s) (Table 1 and Figure 9). These results indicate that EP removal from SSF favours iron release from the protein shell to a large extent. The initial rate of iron release from both WT and EP-deleted SSF at pH 5.8 is smaller than that at pH 6.0, which may be due to larger protein aggregation occurring at pH 5.8 compared with at pH 6.0 (results not shown).

Table 1
Initial rate of iron release from WT holoSSF and holoSSF with the EP deleted, induced by ascorbate (1 mM) at different pH values

Values are means±S.D. (n=3).

 Initial rate of iron release (μM/s) 
Samples pH 5.8 pH 6.0 pH 6.5 
WT holoSSF 0.0569±0.0045 0.194±0.013 0.145±0.023 
EP-deleted holoSSF 0.608±0.063 2.36±0.15 1.787±0.112 
 Initial rate of iron release (μM/s) 
Samples pH 5.8 pH 6.0 pH 6.5 
WT holoSSF 0.0569±0.0045 0.194±0.013 0.145±0.023 
EP-deleted holoSSF 0.608±0.063 2.36±0.15 1.787±0.112 

Comparison of iron release from WT holoSSF and holoSSF with the EP deleted in the presence of ascorbic acid at pH 6.0

Figure 9
Comparison of iron release from WT holoSSF and holoSSF with the EP deleted in the presence of ascorbic acid at pH 6.0

Conditions: 0.2 μM holoSSF or the holoSSF with the EP deleted, 1 mM ascorbic acid, 100 μM ferrozine, in 100 mM Mes and 100 mM NaCl (pH 6.0, 25 °C).

Figure 9
Comparison of iron release from WT holoSSF and holoSSF with the EP deleted in the presence of ascorbic acid at pH 6.0

Conditions: 0.2 μM holoSSF or the holoSSF with the EP deleted, 1 mM ascorbic acid, 100 μM ferrozine, in 100 mM Mes and 100 mM NaCl (pH 6.0, 25 °C).

DISCUSSION

Phytoferritin is unique among all known ferritins in that it contains a specific EP domain at its N-terminal sequence that could impart special properties to the protein [1,6,10]. Although it has been almost 20 years since EP was first reported [10], its biological role remains unknown. Until recently, the EP was reported to serve as a second centre responsible for iron binding and oxidation, at high Fe2+ flux into phytoferritin, through a novel pathway by which the iron is ultimately deposited in the mineral core [12]. However, there is little information available on its function during seedling germination and growth.

The present study demonstrates that WT SSF degradation mainly occurs due to the serine protease-like activity of EP-1 on the outer surface of protein, and that such degradation facilitates iron release from the protein. Initially, WT holoSSF degradation took place during storage (Figure 1). Consistent with this observation, previous reports showed that degraded holoSSF can be directly separated during seed germination, and the amount of degraded holoPSF rose with increasing soaking time [7,14]. The same phenomenon was also observed previously by another group, showing that a 22 kDa subunit was obtained after the soya bean seeds had been soaked long enough to induce germination [30]. Thus phytoferritin degradation seems to constantly take place in vitro and in vivo.

More importantly, the reason why SSF autodegradation occurs during seed germination or storage in vitro was addressed in the present study. Previous studies found that the production of hydroxyl radical (HO) during iron release from holoPSF in the presence of ascorbate can be completely inhibited by iron chelators such as o-phenanthroline and desferrioxamine B. Based on this observation, it was proposed that holoPSF degradation is due to damage by the iron-induced hydroxyl radical [14,15]. However, the fact that protein degradation comes from pure SSF alone (Figure 1), and that the addition of either o-phenanthroline or desferrioxamine B to WT SSF has no effect on such degradation (Supplementary Figure S7 at http://www.BiochemJ.org/bj/427/bj4270313add.htm), has raised uncertainties concerning this proposal. If the observed protein degradation is attributed to the iron-induced HO, it is not understood why no degradation occurs with EP-deleted holoSSF (Figure 4, lane 3) containing the same amount of iron within a protein shell as WT holoSSF (results not shown). The finding that protein degradation occurs with WT apoSSF (Figure 4, lane 2) precludes the possibility that protein degradation stems from iron-induced HO. Thus the protein shell, not iron, is responsible for protein degradation, specifically holoSSF autodegradation. As expected, upon EP removal from holoSSF, the resultant protein (consisting of 23.5 and 21.0 kDa subunits with a molecular mass of ~440 kDa) becomes stable with no degradation under the same experimental conditions (Figure 4, lanes 3 and 4), demonstrating that SSF autodegradation actually corresponds to EP degradation. Thus the EP is a protease-susceptible sequence, possibly because of its exposure on the outer surface of the protein as predicated by the three-dimensional model [6]. However, except for the EP, the other part of SSF seems to be resistant to proteolysis, as suggested by no further degradation in SSF (Figure 1). Consistent with the present observation, it was found that subtilisin can only catalyse a slight hydrolysis of SSF, independent of reaction time [30].

As a typical inhibitor of serine proteases, PMSF is capable of preventing EP degradation, whereas other kinds of protease inhibitors such as EDTA, pepstatin, benzamidine and leupeptin are not (Figure 5). This suggests that EP degradation originates from its serine protease-like activity. Upon treatment of SSF or EP-1 with PMSF, hydroxy groups located on the side chain of serine residues in the EP were deactivated by its integration with hydrogen fluoride (Figures 5 and 7), which was generated from PMSF hydrolysis due to its instability in aqueous solutions at pH ~7.0 [29]. This eliminates the serine protease-like activity of the EP, ultimately inhibiting EP degradation. Agreeing with this idea, two other serine protease inhibitors, containing AEBSF and antipain, also have such an ability (Supplementary Figure S3). Direct evidence of this view comes from protease activity measurements showing that either EP-1 or EP-1 plus EP-2 exhibit significant hydrolysing activity against two peptide–MCA substrates, Boc-Gln-Ala-Arg-MCA (Figure 6A) and N-succinyl-Ala-Phe-Lys-MCA (Figure 6B). In comparison, hydrolysing activity of the EP-2 alone is much weaker. Thus EP-1 is mainly responsible for SSF degradation. The large difference between the hydrolysing activity of EP-1 and EP-2 may reside in their somewhat dissimilar amino acid residues. EP-1 contains two serine residues at positions 2 and 20, whereas EP-2 has only one serine residue at position 2 (Figure 3A). The above conclusion was further confirmed by a comparison of rH-1 and rH-2 stability, showing that rH-2 is a stable protein molecule whereas its analogue, rH-1, is susceptible to degradation (Figure 8). These results clearly demonstrate that WT SSF autodegradation during storage mainly stemmed from the serine protease-like activity of EP-1. Consistent with the present observation, previous studies showed that the H-2 subunit was more resistant to proteolysis compared with the H-1 subunit [8]. Further support comes from a recent report showing that recombinant His6–H-1 ferritin (in which each subunit contains six histidine residues at its N-terminal for convenient purification) is prone to degradation, whereas its analogue, recombinant His6–H-2 ferritin, is stable [31].

What advantage might EP autodegradation render to phytoferritin during seedling germination and growth? The answer to this question may lie in the experimental results of iron reductive release from WT and EP-deleted holoSSF (Figure 9 and Table 1). The rate of iron release from EP-deleted holoSSF is pronouncedly larger than that from WT holoSSF at different pH values (5.8, 6.0 and 6.5) (Table 1). Thus EP deletion facilitates iron release from ferritin. In support of this conclusion, it has been established that the majority of iron in soya bean seed was stored within ferritin [3,32] to meet the demand of various key physiological processes, such as electron transfer and DNA synthesis in the seedling stage [1,3]. It was visualized that two-thirds of iron in soya bean seeds move from bean cotyledons to the seedling axis during the first week of germination [33]. Thus fast iron release from ferritin might be beneficial to these processes. Alternatively, phytoferritin is not only an iron depot, but a phosphorus storage molecule with a iron/phosphorus ratio of <3:1 (results not shown), in accordance with previous measurements [1,11]. Accompanied by iron release, phosphate ions were therefore also liberated more quickly from EP-deleted holoSSF than its analogue, again favouring the above physiological processes [14].

The crystal structure analysis of the soya bean ferritin indicates that the EP stabilizes the protein shell by maintaining the conformation of the C-helix [4]. Therefore it is likely that removal of the EP domain destabilizes the outer case of the ferritin structure, leading to the facile release of iron and phosphorus. This is reminiscent of the mechanism of iron release from animal ferritin which is usually a cytosolic molecule. Many studies have shown that cytosolic ferritin gains entry into lysosomes by autophagy, and that ferritin degradation within lysosomes is responsible for iron release [34,35]. Thus there is something in common between the mechanism of iron release from plant and animal ferritins, namely, their degradation occurs prior to iron exit from protein. However, phytoferritin degradation is slightly partial degradation. As a result, the degraded phytoferritin might retain an intact protein shell, as suggested by the result showing that it contains the same amount of iron within a protein shell as WT holoSSF (results not shown). Thus reductants, such as ascorbate, are still required for reducing Fe3+ caged in phytoferritin, and resultant ferrous ions diffuse out of ferritin shell plant ferritin through 3- or 4-fold channels [13]. In contrast, complete degradation of animal ferritin by hydrolases existing in the lysosomes results in iron release from protein without participation of the reductants [34,35].

In conclusion, the present study demonstrates that the EP on SSF possesses a serine protease-like activity previously unrecognized, thus resulting in its autodegradation. Associated with the degradation is faster iron release from ferritin to meet the requirement of seedling growth, representing a novel pathway for how phytoferritin controls its iron complement through a specific domain in phytoferritin, the EP. Alternatively, since SSF is compartmentalized in amyloplasts, one of the most vital and yet least-understood organelles in seedling germination and early growth [28], the detailed knowledge of ferritin shown in the present study is beneficial to understanding the function of this organelle.

Abbreviations

     
  • AEBSF

    4-(2-aminoethyl)benzenesulfonyl fluoride

  •  
  • AMC

    7-amino-4-methylcoumarin

  •  
  • Boc

    t-butoxycarbonyl

  •  
  • EP

    extension peptide

  •  
  • MALDI–TOF-MS

    matrix-assisted laser-desorption ionization–time-of-flight MS

  •  
  • MCA

    (7-methoxycoumarin-4-yl)acetyl

  •  
  • PSF

    pea seed ferritin

  •  
  • rH-1

    (rH-2), recombinant H-1 (H-2) ferritin

  •  
  • SSF

    soya bean seed ferritin

  •  
  • TP

    transit peptide

  •  
  • WT

    wild-type

AUTHOR CONTRIBUTION

Xiaoping Fu performed the biochemical analysis, mass spectra and contributed to manuscript writing. Jianjun Deng contributed to prepration of amyloplasts and ferritin, and writing. Haixia Yang performed CD and fluorescence spectroscopy. Taro Masuda. Fumiyuki Goto and Toshihiro Yoshihara contributed to preparation of rH-1 and rH-2. Guanghua Zhao contributed to experimental design, ideas and manuscript writing.

FUNDING

This work was supported by the National Natural Science Foundation of China [grant number 30972045]; the Chinese Universities Scientific Fund [grant number 2009-3-10]; and the China High-Tech (863) project [grant number 2007AA10Z333].

References

References
1
Harrison
P. M.
Arosio
P.
Ferritins: molecular properties, iron storage function and cellular regulation
Biochim. Biophys. Acta
1996
, vol. 
1275
 (pg. 
161
-
203
)
2
Chasteen
N. D.
Harrison
P. M.
Mineralization in ferritin: an efficient means of iron storage
J. Struct. Biol.
1999
, vol. 
126
 (pg. 
182
-
194
)
3
Theil
E. C.
Iron, ferritin, and nutrition
Annu. Rev. Nutr.
2004
, vol. 
24
 (pg. 
327
-
343
)
4
Masuda
T.
Goto
F.
Yoshihara
T.
Mikami
B.
Crystal structure of plant ferritin reveals a novel metal binding site that functions as a transit site for metal transfer in ferritin
J. Biol. Chem.
2010
, vol. 
285
 (pg. 
4049
-
4059
)
5
Crichton
R. R.
Herbas
A.
Chavez-Alba
O.
Roland
F.
Identification of catalytic residues involved in iron uptake by L-chain ferritins
J. Biol. Inorg. Chem.
1996
, vol. 
1
 (pg. 
567
-
574
)
6
Lobréaux
S.
Yewdall
S. J.
Briat
J. F.
Harrison
P. M.
Amino-acid sequence and predicted three-dimensional structure of pea seed (Pisum sativum) ferritin
Biochem. J.
1992
, vol. 
288
 (pg. 
931
-
939
)
7
Van Wuytswinkel
O.
Savino
G.
Briat
J. F.
Purification and characterization of recombinant pea-seed ferritins expressed in Escherichia coli: influence of N-terminus deletions on protein solubility and core formation in vitro
Biochem. J.
1995
, vol. 
305
 (pg. 
253
-
261
)
8
Masuda
T.
Goto
F.
Yoshihara
T.
A novel plant ferritin subunit from soybean that is related to a mechanism in iron release
J. Biol. Chem.
2001
, vol. 
276
 (pg. 
19575
-
19579
)
9
Li
C.
Hu
X.
Zhao
G.
Two different H-type subunits from pea seed (Pisum sativum) ferritin that are responsible for fast Fe(II) oxidation
Biochimie
2009
, vol. 
91
 (pg. 
230
-
239
)
10
Ragland
M.
Briat
J. F.
Gagnon
J.
Laulhere
J. P.
Massenet
O.
Theil
E. C.
Evidence for conservation of ferritin sequences among plants and animals and for a transit peptide in soybean
J. Biol. Chem.
1990
, vol. 
265
 (pg. 
18339
-
18344
)
11
Waldo
G. S.
Wright
E.
Wang
Z. H.
Briat
J. F.
Theil
E. C.
Sayers
D. E.
Formation of the ferritin iron mineral occurs in plastids: an x-ray absorption spectroscopy (EXAFS) study
Plant Physiol.
1995
, vol. 
109
 (pg. 
797
-
802
)
12
Li
C.
Fu
X.
Qi
X.
Hu
X.
Chasteen
N. D.
Zhao
G.
Protein association and disassociation regulated by ferric ion. A novel pathway for oxidative deposition of iron in pea seed ferritin
J. Biol. Chem.
2009
, vol. 
284
 (pg. 
16743
-
16751
)
13
Ravet
K.
Touraine
B.
Boucherez
J.
Briat
J. F.
Gaymard
F.
Cellier
F.
Ferritins control interaction between iron homeostasis and oxidative stress in Arabidopsis
Plant J.
2009
, vol. 
57
 (pg. 
400
-
412
)
14
Lobréaux
S.
Briat
J. F.
Ferritin accumulation and degradation in different organs of pea (Pisum sativum) during development
Biochem. J.
1991
, vol. 
274
 (pg. 
601
-
606
)
15
Laulhere
J. P.
Laboure
A. M.
Briat
J. F.
Mechanism of the transition from plant ferritin to phytosiderin
J. Biol. Chem.
1989
, vol. 
264
 (pg. 
3629
-
3635
)
16
Masuda
T.
Goto
F.
Yoshihara
T.
Ezure
T.
Takashi
S.
Kobayashi
S.
Shikata
M.
Utsumi
S.
Construction of homo- and heteropolymers of plant ferritin subunits using an in vitro protein expression system
Protein Expr. Purif.
2007
, vol. 
56
 (pg. 
237
-
246
)
17
Bauminger
E. R.
Harrison
P. M.
Hechel
D.
Nowik
I.
Treffry
A.
Mössbauer spectroscopic investigation of structure–function relations in ferritins
Biochim. Biophys. Acta
1991
, vol. 
1118
 (pg. 
48
-
58
)
18
Treffry
A.
Hirzmann
J.
Yewdall
S. J.
Harrison
P. M.
Mechanism of catalysis of Fe(II) oxidation by ferritin H-chains
FEBS Lett.
1992
, vol. 
302
 (pg. 
108
-
112
)
19
Laemmli
U. K.
Cleavage of structural proteins during the assembly of the head of bacteriophage T4
Nature
1970
, vol. 
227
 (pg. 
680
-
685
)
20
Nishikata
M.
Trypsin-like protease from soybean seeds. Purification and some properties
J. Biochem.
1984
, vol. 
95
 (pg. 
1169
-
1177
)
21
Guo
Z. J.
Lamb
C.
Dixon
R. A.
A serine protease from suspension-cultured soybean cell
Phytochemistry
1998
, vol. 
47
 (pg. 
547
-
553
)
22
Morita
T.
Kato
H.
Iwanaga
S.
Takada
K.
Kimura
T.
Sakakibara
S.
Fluorogenic substrates for α-thrombin, factor Xa, kallikreins, and urokinase
J. Biochem.
1977
, vol. 
82
 (pg. 
1495
-
1498
)
23
Hynes
M. J.
Coinceanainn
M.Ó.
Investigation of the release of iron from ferritin by naturally occurring antioxidants
J. Inorg. Biochem.
2002
, vol. 
90
 (pg. 
18
-
21
)
24
Tonomura
B.
Nakatani
H.
Ohnishi
M.
Yamaguchi-Ito
J.
Hiromi
K.
Test reactions for a stopped-flow apparatus. Reduction of 2, 6-dichlorophenolindophenol and potassium ferricyanide by L-ascorbic acid
Anal. Biochem.
1978
, vol. 
84
 (pg. 
370
-
383
)
25
Stookey
L. L.
Ferrozine-a new spectrophotometric reagent for iron
Anal. Chem.
1970
, vol. 
42
 (pg. 
779
-
781
)
26
Zhao
G.
Bou-Abdallah
F.
Arosio
P.
Levi
S.
Janus-Chandler
C.
Chasteen
N. D.
Multiple pathways for mineral core formation in mammalian apoferritin. The role of hydrogen peroxide
Biochemistry
2003
, vol. 
42
 (pg. 
3142
-
3150
)
27
Denyer
K.
Smith
A. M.
The capacity of plastids from developing pea cotyledons to synthesise acetyl CoA
Planta
1988
, vol. 
173
 (pg. 
172
-
182
)
28
Yu
Y.
Mu
H. H.
Mu-Forster
C.
Wasserman
B. P.
Polypeptides of the maize amyloplast stroma: stromal localization of starch biosynthetic enzymes and identification of an 81-kilodalton amyloplast stromal heat shock cognate
Plant Physiol.
1998
, vol. 
116
 (pg. 
1451
-
1460
)
29
Fahrney
D. E.
Gold
A. M.
On the problem of the serine-histidine hydrogen bond in the active site of α-chymotrypsin
J. Am. Chem. Soc.
1963
, vol. 
85
 (pg. 
349
-
350
)
30
Sczekan
S. R.
Joshi
J. G.
Isolation and characterization of ferritin from soyabeans (Glycine max)
J. Biol. Chem.
1987
, vol. 
262
 (pg. 
13780
-
13788
)
31
Xiangbai
D.
Tang
B.
Li
J.
Xu
Q.
Fang
S.
Hua
Z.
Expression and purification of intact and functional soybean (Glycine max) seed ferritin complex in Escherichia coli
J. Microbiol. Biotechnol.
2008
, vol. 
18
 (pg. 
299
-
307
)
32
Ambe
S.
Ambe
F.
Nozaki
T.
Mössbauer study of iron in soybean seeds
J. Agric. Food Chem.
1987
, vol. 
35
 (pg. 
292
-
296
)
33
Biddulph
O.
Truog
E.
The translocation of minerals in plants
Mineral Nutrition in Plants
1951
Madison
University of Wisconsin Press
(pg. 
261
-
263
)
34
Kurz
T.
Terman
A.
Gustafsson
B.
Brunk
U. T.
Lysosomes and oxidative stress in aging and apoptosis
Biochim. Biophys. Acta
2008
, vol. 
1780
 (pg. 
1291
-
1303
)
35
Radisky
D. C.
Kaplan
J.
Iron in cytosolic ferritin can be recycled through lysosomal degradation in human fibroblasts
Biochem. J.
1998
, vol. 
336
 (pg. 
201
-
205
)

Supplementary data