Screening and structural and functional investigation of a novel ferritin from Phascolosoma esculenta

Abstract

Ferritins are primary iron storage proteins and play a crucial role in iron storage and detoxification. Yeast two‐hybrid method was employed to screen the cDNA library of Phascolosoma esculenta. Sequence of positive colony FER147 was analyzed. The higher similarity and conserved motifs for ferritin indicated that it belonged to a new member of ferritin family. The interaction between Ferritin and Fer147 was further confirmed through co‐immunoprecipitation. The pET‐28a‐FER147 prokaryotic expression vector was constructed. The expressed recombinant Fer147 was then isolated, purified, and refolded. When ferritins were treated by different heavy metals, several detection methods, including scanning electron microscopy (SEM), circular dichroism (CD), and inductively coupled plasma–mass spectrometry (ICP‐MS) were applied to examine the structures and functions of the new protein Fer147, recombinant P. esculenta ferritin (Rferritin), and natural horse‐spleen ferritin (Hferritin). SEM revealed that the three ferritin aggregates changed obviously after different heavy metals treatment, meanwhile, a little different in aggregates were detected when the ferritins were trapped by the same heavy metal. Hence, changes in aggregation structure of the three proteins are related to the nature of the different heavy metals and the interaction between the heavy metals and the three ferritins. CD data suggested that the secondary structure of the three ferritins hardly changed after different heavy metals were trapped. ICP–MS revealed that the ferritins exhibit different enrichment capacities for various heavy metals. In particular, the enrichment capacity of the recombinant Fer147 and Rferritin is much higher than that of hferritin.

Introduction

Iron is implicated in normal cell growth and considered as an essential element for all living organisms because this element plays a key role in several important biological processes,1 such as oxygen transport, electron transfer, and enzymatic reactions.2 However, excess iron is potentially toxic because it can participate in redox reactions that may induce the formation of reactive oxygen species and free iron, along with superoxide anion and hydrogen peroxide. These components provide a deadly mixture that can stimulate the production of hydroxyl radicals, which are extremely reactive and thus cause lipid oxidation.3 Oxidative stress and lipid oxidation are related to several pathologies.4 Therefore, iron must be closely controlled, and cells must maintain a transient pool of readily available iron within a narrow range to provide enough iron for cell functions and to limit any excess iron that may induce toxic reactions. The iron not utilized in metabolism is stored in the iron‐binding protein ferritin, which has a protein shell composed of 24 subunits arranged in a 4–3–2 symmetry.5 Results have showed that natural ferritin could store multiple toxic metal ions by its great storage capacity,6 resulting to detoxification of heavy metal in vivo.

Phascolosoma esculenta, is a Sipuncula only distributed in China. Ferritin plays an important role in its strong adaptability to harsh living environments, especially to an environment with high iron content.7 Studies on ferritins have focused on the regulation of iron metabolism, immunosensing, and inducing conditions of ferritin expression. However, in vitro or in vivo experiments regarding this protein have yet to be performed despite its significance in the protection and maintenance of life. We found a novel protein interacting with ferritin which was named Fer147 and a number of ferritin superfamily by a yeast two‐hybrid system. To further elucidate Fer147′s function on heavy metal detoxification compared with P. esculenta ferritin, Fer147 was expressed in Escherichia coli BL21 strain. Scanning electron microscopy (SEM) and circular dichroism (CD) were employed to observe the differences in aggregation characteristics and secondary structures among the new ferritin (Fer147), recombinant P. esculenta ferritin (Rferfritin) and standard horse spleen ferritin (Hferritin). Inductively coupled plasma–mass spectrometry (ICP–MS) was also conducted to determine the chelating capacity for different heavy metals of the three ferritins.

Results

Identification of ferritin‐interacting proteins by yeast two‐hybrid system

To identify Ferritin‐associated proteins, we first screened a cDNA library from P. esculenta cells using the yeast two‐hybrid system with the ORF sequence of FERRITIN and the GAL4 DNA binding domain as bait. When the bait vector pGB‐FERRITIN and testis P. esculenta cDNA library were co‐transformed into yeast Y190, 180 positive colonies were obtained in the SD/‐Trp/‐Leu/‐His/3‐AT medium, after β‐galactosidase activity assay, all of them were showed positive colonies. The plasmids from 30 of them randomly selected were extracted, transformed into competent E.coli DH5α and incubated in LB‐ampicillin medium. The amplified plasmids were co‐transformed with pGB‐FERRITIN into yeast Y190, 25 colonies were confirmed positive by β‐galactosidase activity assay and sequenced. Nucleotide sequencing and a BLAST search revealed that 23 positive clones encoded ferritins themselves, one positive clone encoded Hemerythrin and one encoded Fer147.

Primary structure of Fer147

The cDNA of Fer147 contained a DNA fragment of 916 bp. Sequence analysis revealed no stop codon in the region between the 5′‐end and the first Met codon at 174 bp (Fig. 1). The first stop codon was found at 695 bp. Initially, we were concerned that the 5′‐end of the cDNA clone did not contain the actual translation‐initiation code of the gene. To test this possibility, we performed 5′‐RACE using two different cDNA fragments by screening the library by PCR, but no extended sequence was observed in the 5′‐end regions of these fragments. In addition, the size of the cDNA (916 bp) was close to the size of the P. esculenta ferritin reported by BLAST. Given these results, we tentatively propose that this clone is a full‐length cDNA, and the Met codon at 174 bp was the true translation‐initiation codon. The ORF encodes a polypeptide of 174 amino acids.

Figure 1.

Structure of the novel cDNA. The base sequence and the encoded amino acid sequence of fer147 are depicted and compared with P. esculenta ferritin. Note: (A.) ferroxidase center, which belongs to H‐Ferritin in vertebrate contain seven conserved residues (Glu‐25, Tyr‐32, Glu‐59, Glu‐60, His‐63, Glu‐105, Gln‐139), and all the seven residues of the two ferritins are the same. ( $\mathrm{A}$:) iron‐ion channel, which are common in both mammalian H‐ and L‐ferritins and contain three conserved residues (Lys‐106, Asp‐129, Glu‐132), and all the three residues of the two ferritins are the same. (A⁰) Iron nucleation center, which belongs to L‐ferritins consists of four conserved residues (Fer147: Lys‐55, Ser‐58, Glu‐59, Gln‐62; Rferritin:Lys‐55, Ser‐58, Glu‐59, Glu‐62). (A) iron‐binding‐region signature (from Glu‐59 to Arg‐77); (A) N‐glycosylation site

BLAST analysis revealed that Fer147 exhibited fairly high similarities with the P. esculenta ferritin (81%). The key feature of the ferritins is their ability to oxidize ferrous iron to the ferric state during the iron‐core assembly process. This “ferroxidation” capacity is endowed by the presence of active sites, known as the “ferroxidase centers.” These sites are located in the middle of the four‐helical bundle of individual subunit.8 Ferritin cages without ferroxidase centers were shown to take up iron, albeit slowly.9 Fer147 contains the same ferroxidase center, iron ion channel, and N‐glycosylation site with P. esculenta ferritin. The similarities in the ferrihydrite nucleation center and putative iron‐binding region signature between Fer147 and Rferritin were 75% and 74%, respectively. This finding suggests that the encoded protein is a potentially novel subunit of the P. esculenta ferritin. In most vertebrates, the ferritins are known to comprise different proportions of two subunit types, termed H‐ and L‐chains. The two subunits share approximately 55% of their sequence as well as their multi‐helical three‐dimensional structure. Previously, in invertebrates, ferritin was thought to contain only one type of subunit, which is functionally equivalent to the two subunits of ferritin in vertebrates. However, considering our results, we can conclude the possibility of two ferritin subunits in invertebrates, similar to vertebrates, and both subunits contain the ferroxidase centers. In accordance with the differences in structure and function among mammalian ferritins, plant ferritins, and bacterioferritins, we propose that Rferritin and Fer147 play roles in iron storage in different condition.

Fer147 interacts with ferritin

Through the yeast two‐hybrid system, we demonstrated that Fer147 could interact with ferritin. To directly examine protein–protein interactions in the cells, 293t cells were transfected with Myc‐ferritin and HA‐fer147. The cell lysates were subjected to immunoprecipitation with anti‐Myc antibody, followed by Western blot analysis using the anti‐HA antibody. The interaction between ferritin and fer147 was detected in co‐IP experiments (Fig. 2).

Figure 2.

The interaction between fer147 and ferritin was confirmed by co‐IP experiments. 293T cells were co‐transfected with Fer147‐HA and Ferritin‐myc. After 72 h whole cell lysates were subject to immunoprecipitation (IP) with anti‐myc antibodies and co‐immunoprecipitation analyzed by Western Blot (WB) using anti‐HA antibodies. WB of HA‐tagged Fer147 and Ferritin‐myc from 2% input is shown as positive control. IgG, immunoglobulin used for negative control

Expression and purification of recombinant fer147

The proteins extracted from the bacteria carrying the recombinant plasmid PET‐28a‐Fer147 were separated by SDS–PAGE, and a single protein band between 20.1 and 29 kDa was visualized. The protein was slightly larger than the native protein (about 21 kDa) because of the addition of N‐terminal six‐his tag in the reconstructed sequence expressed from the expression vector. Figure 3 shows that the amount of the expressed fer147 increased significantly with IPTG induction time. The optimal condition for protein expression was 1.0 mmol/L IPTG at 37°C for 5 h [Fig. 3(A)]. Likewise, purified Fer147 were resolved by SDS–PAGE and Coomassie brilliant blue staining showed a single band between 20.1 and 29 kDa [Fig. 3(B)].

Figure 3.

Expression and purification of recombinant protein Fer147 in E. coli BL21. (A) Lane M: low molecular marker; Lane 1–6: total protein of PET‐28a‐Fer147 induced by IPTG for 1, 2, 3, 4, 5, and 6 h, respectively. Lane 7: total protein of non‐induced PET‐28a; Lane 8: total protein of PET‐28a induced by IPTG for 5 h; Lane 9: total protein of non‐induced PET‐28a‐Fer147. (B) Lane Fer147: purified recombinant protein of PET‐28a‐Fer147; Lane M: molecular marker

Molecular mass of Fer147 after renaturation

To determine the aggregation state of fer147 after renaturation, Size‐exclusion chromatography (SEC) was performed, and Hferritin was used as a control. The peak of Hferritin eluted in the exclusion volume of the column indicated a molecular mass about 500 kDa. Figure 4 showed that the molecular mass of Fer147 was also about 500 kDa and both proteins eluted in a filtration column at a position consistent with the 24‐meric form of ferritin.

Figure 4.

Elution profile of Hferritin and Fer147 from a superdex‐200 column. Size exclusion chromatography was performed with affinity‐purified, recombinant Fer147 (green curve) and Hferritin (black curve) using a preparative S‐200 Superdex column

Heavy‐metal‐chelating ability

Information obtained from SEM

The self‐aggregation of proteins often generates multilayer films and irregular dense clusters throughout a surface when diluted protein solutions are deposited and dried on surfaces under natural conditions. F. Akif Tezcan10 indicated that the use of high protein and Cu concentrations produces considerable amounts of higher‐order aggregates. In our paper, Hferritin, Rferritin and Fer147 without heavy metals treatment were used as control in SEM experiment. The morphology of samples which the protein is absent was shown as Supporting Information Figure S1. Figure 5 shows that the three ferritin proteins exhibited a little difference in morphology after the same heavy metal treatment; and the same ferritin showed obviously different morphology after enriching different heavy metals. Fer147 without heavy metals exhibited a regular spherical shape, with spherical protein complexes. Hferritins without heavy metals formed complexes of regular size and shapes, and linked together to smooth lump‐like aggregates. In the absence of heavy metals, Rferritin formed a spherical shape and was well dispersed, with a diameter of 0.5 μm. After treatment with Mn²⁺, the three ferritins all formed large particles with similar flower‐like structures, but with different sizes and folds. The sizes of Hferritin–Mn, Fer147–Mn and Rferritin–Mn were about 3, 9, and 6 μm, respectively. Cd²⁺‐enriched the three ferritins formed similar aggregation, while Rferritin‐Cd and Fer147‐Cd formed small dendritic structures with different sizes and aggregated into stacked layers. The size of Rferritin–Cd was about 0.3 μm in diameter, which was slightly bigger than that of Fer147–Cd. Cd²⁺‐enriched Hferritins formed compact aggregates. The Zn²⁺‐enriched Hferritin and Rferritin generated numerous dispersed small dots and did not aggregate with each other; the Zn²⁺‐enriched Fer147 also showed many small dots, but these dots aggregated with one another to form a nebula‐like structure. The Cu²⁺‐enriched Hferritin stacked up to the aggregation with irregular granules, and the Cu²⁺‐enriched Fer147 showed loose laminated structure with small particles, whereas Rferritin–Cu exhibited a compact laminated structure. The Fe²⁺‐treated Hferritin aggregates formed regular blocks with smooth surfaces and diagonal lengths arranged from 0.2–0.5 μm, both Fe²⁺‐enriched Fer147 and Rferritin formed cave‐like structures with very small particles. Three Pb²⁺‐enriched ferritins aggregates were composed of many pellets about 0.3 μm in diameter, but the aggregate shapes were different among them. The Pb²⁺‐enriched Hferritin formed a big solid structure and the Pb²⁺‐enriched Fer147 formed a relatively decentralized structure, whereas Pb²⁺‐enriched Rferritin formed a loose group structure, but the three aggregation shapes were all composed of many pellets. The Hg²⁺‐enriched Hferritin formed a branched structure with smooth surface, whereas both Hg²⁺‐enriched Fer147 and Rferritin showed a small and relatively scattered flower‐like structure with a wrinkled surface but with different sizes of approximately 1.5 and 0.8 μm, respectively. The Cr³⁺‐enriched Fer147 and Rferritin aggregates were composed of numerous very small pellets (0.1 μm in diameter), which gathered together to form a three‐dimensional structure, whereas Hferritin–Cr was composed of these pellets to form a single‐layer structure.

Figure 5.

SEM images revealed the differences between the three ferritin proteins treated with the same heavy metals. Hferritin, Fer147 and Rferritin showed the aggregates of the three ferritins without any heavy metals treatments. H(R)ferritin (Fer147)‐heavy metals represent the aggregates morphology of Hferritin, Rferritin or Fer147 treated by the corresponding heavy metals

Information obtained from CD

CD spectrometry revealed characteristics of the three ferritins treated with different heavy metals; it provided information on whether the molecular structure of ferritins had greatly changed with treatment. The secondary‐structure contents of the three ferritins were estimated from the CD data. As shown in Figure 6, the CD curves of all Hferritins [Fig. 6(A)] and Rferritin [Fig. 6(B)] treated with different heavy metals gave similar shape along with wavelength from 190 nm to 250 nm, which suggested that the secondary structures of all of the Hferritins and Rferritins were highly similar. They showed the typical CD spectral characteristics of α‐helix with two clear negative peaks at 208 and 222 nm and a positive band close to 195 nm. Thus the secondary structure of Hferritins and Rferritin treated with different heavy metals mainly comprises the α‐helix which indicated that the heavy metals did not change the secondary structure of Hferritin and Rferritin significantly. For Fer147 [Fig. 6(C)], the CD spectra of Cr³⁺, Pb²⁺, Hg²⁺ treatment samples showed a positive peak at 198 nm which differed from the typical spectra of helical proteins and a much higher protein concentration or the presence of heavy metals ions may account for the unusual spectra. The CD spectra of the other heavy metals treatment fer147 showed typical α‐helix CD spectra. The secondary structural estimation for the three ferritins with and without heavy metals treatment are given in Supporting Information Tables S1–S3. Considering the CD data analysis, we can conclude that all of the metals tested in this study can hardly alter the secondary structures of the three ferritins.

Figure 6.

CD spectra of rferritin, fer147, and hferritin treated with different heavy metals. (A) Far‐UV CD spectra of Hferritin without heavy metal, and in the presence of Fe (II), Pb (II), Cd (II), Cr (III), Cu (II), Mn (II), Hg (II), and Zn (II); (B): Far‐UV CD spectra of Rferritin with and without these heavy metals treatment; (C): Far‐UV CD spectra of Fer147 with and without these heavy metals treatment

Heavy metal uptake capacity of the three ferritins

Figure 7 presents the heavy metal concentrations from the different treatment groups by ICP–MS. The metal content was considerably higher in the case of the recombinant ferritins (Fer147 and Rferritin) compared with Hferritin (P < 0.01). The difference in heavy metal uptake capacity among the three ferritins treated by the same heavy metals was statistically significant; even the same ferritin treated by different heavy metals showed different heavy metal uptake capacities. Hferritins, Fer147, and Rferritin all could enrich higher amounts of Fe than the other heavy metals. Only 535 Fe ions in the iron core per molecule of Hferritin were detected, a number significantly lower than those of Fer147 (2000 irons per molecule) and Rferritin (1200 irons per molecule). Approximately 4500 iron atoms can be theoretically stored as the mineral ferrihydrite.11 However, A. Lewin12 found that isolated proteins generally contain much less iron atoms than the theoretical content which was consistent with our ICP‐MS results. The relative capacity of Hferritin to enrich different heavy metals is as follows: Fe²⁺ > Hg²⁺ > Mn²⁺ > Cd²⁺ > Cr³⁺ > Pb²⁺ > Cu²⁺ > Zn²⁺. No obvious differences were found among the Hferrritins treated by different heavy metals except for the Fe‐treated hferritin. The relative capacity of Fer147 to enrich different types of heavy metals is as follows: Fe²⁺ > Hg²⁺ > Cd²⁺ > Cr³⁺ > Mn²⁺ > Pb²⁺ > Cu²⁺ > Zn²⁺. The relative uptake capacity of Rferritin for different types of heavy metals is as follows: Fe²⁺ > Mn²⁺ > Hg²⁺ > Cd²⁺ > Cr³⁺ > Zn²⁺ > Cu²⁺ > Pb²⁺. We can see from ICP‐MS data that the enriching capacity of Recombinant Fer147 and Rferritin was higher than that of Hferritin, meanwhile, the enriching capacity of Fer147 for more toxic heavy metals was more higher than that of Rferritin.

Figure 7.

Heavy metal concentrations in different ferritin treatment groups. Samples were prepared by digestions of the material in a small scale, and analyzed as described in Material and methods section. For each series, two independent experiments performed, values are shown as means ± SD calculated from the independent results. Concentrations are based on dry‐weight. Significant differences between these samples were indicated by letters. Different letters represent significant difference (P < 0.01)

Discussion

Many proteins contain irons either within their own structures or bound to their active sites. These iron‐containing proteins are involved in numerous biological processes; some of these proteins serve as biomarkers of clinical pathologies not only related to iron homeostasis but also to other physiological disorders.13 Ferritin is considered as a major iron‐containing protein and important antioxidant because it sequesters iron and exhibits a ferroxidase function; thus, iron is stored in a ferric form without inducing Fenton reactions.14 The ferritin shell with a cavity structure is mostly piled symmetrically by its 24 subunits. This specific molecular structure15 makes ferritin store small molecular compounds, such as various organ phosphorus pesticides and various heavy metals.16 As a major Fe storage protein, ferritin is conserved in most species and permits the storage of up to 4500 Fe (III) atoms, and its Fe can be exchanged with other ions,17 this phenomenon makes ferritin optimal for heavy metals detoxification.

In this study, a novel ferritin protein Fer147 which was a member of ferritin was found from P. esculenta using a yeast two‐hybrid system. Ferritins lack a signal peptide in most species except in arthropods and some mollusks.18 This study showed that Fer147 possessed a unique structural domain of ferritin without a signal peptide; this finding suggested that Fer147 is not a secreted protein. Ferritins in bacteria, plants, and invertebrates are exclusively composed of H‐like subunits. Ferritins with a combination of H and L subunits are found only in vertebrates and have been thoroughly characterized in mammals. We found that Fer147 possessed both the ferroxidase center of the H subunit and iron nucleation sites of the L subunit; this result indicated Fer147 was functionally equivalent to the two subunits of ferritin in vertebrates.19 Then fer147 was expressed by E. coli BL21 strain, several detection technology such as SEM, CD and ICP–MS were used to study the structure and function of Fer147.

The SEM analysis revealed obvious differences between the same ferritins after adsorption of different heavy metals. This finding is consistent with the results reported by Mann,20 who showed significant differences between Hferritins reconstituted with Mn and Fe oxides. There maybe three reasons for differences in aggregation structure among the three ferritins treated by different heavy metals. First, the observed aggregate morphology of the three ferritins treated with different heavy metals could be related to the nature of the metal cores, the dependence of ferritin aggregate structure on the nature of the metal core is suggested by our SEM results that obvious differences occurred among the same ferritin treated by different heavy metals and small differences occurred among the three different ferritins treated by the same heavy metals, this phenomenon may be explained by the different binding sites used by ferritins to bind the different heavy metals, for example, Cd²⁺ was capable of binding to both the inside and outside of Hferritin,21 this may be also the reason why the same ferritin has different abilities to enrich different heavy metals; second, ferritin could retain its primary structure after demetalation, but its magnetism and conductivity would be altered,22 which indicated that different heavy metals in iron core could change the conductivity and magnetism of ferritin, further change the number of aggregation ferritin molecular and the final aggregation structure; last, when ferritins are dried, forces, such as hydrophobic and electrostatic forces, likely pull proteins together into a clustered morphology,23 usually, the stronger the metal ion hydration was, the greater role it played in promoting ferritin aggregation, resulting in the formation of larger ferritin aggregates.24 Results from CD data showed that heavy metals can hardly alter the secondary structures of the three ferritins. This result proved that when exposed to heavy metals, ferritins are chemically stable. This property makes the ferritins promising candidates for heavy metals detoxification.

Data from ICP‐MS showed that Hferritin has the lowest enriching capacity for every heavy metal among the three ferritins. The prevailing mechanism for iron loading suggests that Fe²⁺ binds to the ferritin protein and migrates to an enzyme site, named the ferroxidase center, where oxidation to Fe³⁺ occurs. After oxidation, the iron migrates from the ferroxidase center to the protein interior where it forms the iron mineral.25 The low enriching capacity of Hferritin may be due to the source of Hferritin. The commercially available and most frequently used Hferritin contains approximately 90% L‐subunits.26 Only H‐subunit has highly conserved ferroxidase center27 which could catalyze the oxidation of Fe (II), as lacking the catalytic ferroxidase centre, Hferritin can still oxidize iron but at much lower rates than the two recombinant ferritins which have both ferroxidase centers in H‐subunit and nucleation site in L‐subunit. Additionally, Fer147 had higher enriching capacity for toxic heavy metals such as Cu²⁺, Pb²⁺ and Hg²⁺ than Rferritin had. Therefore, Fer147 may be more proper for implication in heavy metals detoxification and environmental remediation.

In conclusion, the recombinant P. esculenta ferritins, especial Fer147 showed excellent ability to enrich various heavy metals. The uptake process featured high yield, stability, simple to implement and low‐cost. Thus, Fer147 could be a proper bio‐material in the manufacture of drugs for heavy metal detoxification and environmental remediation.

Materials and Methods

Screening of ferritin‐binding proteins by using the yeast two‐hybrid system

Construction of the cDNA library of P. esculenta28

Total RNAs were extracted from P. esculenta cells by using TRIzol reagent (Invitrogen, Shanghai, China), and mRNAs were obtained in accordance with the instructions of FastTrack MAG mRNA isolation kit. Double‐stranded cDNA was synthesized using a CloneMiner II cDNA library construction kit (ThermoFisher, Shanghai, China) in accordance with the manufacturer's instructions. The qualities of the isolated total RNA and mRNA were identified through 1% agarose gel electrophoresis. The PCR products were linked to an attB1 adapter for 16–24 h at 16°C and fractionated using columns to filter the fragments. The fragments larger than 500 bp were collected and ligated to a pGADT7‐DEST vector via recombinant reaction between an attL‐containing entry vector and an attR‐containing destination vector. Then, the ligated product was electrically transformed into E. coli DH10B to construct a cDNA library for use in the yeast two‐hybrid system. A 50 μL sample of the cDNA library mixture was plated onto Luria‐Bertani (LB)–ampicillin plates containing Isopropyl β‐D‐Thiogalactoside (IPTG) and 5‐Bromo‐4‐chloro‐3‐indolyl β‐D‐galactopyranoside (X‐gal) and then incubated overnight at 37°C to determine the quality of the constructed P. esculenta cDNA library. White colonies were counted to evaluate the capacity of the constructed cDNA library. The recombinant rate and average length of the inserted cDNA fragment was determined through PCR amplification.

Construction of bait vector pGB–Ferritin29

We used the plasmid pGB containing the sequences encoding the Gal4 DNA‐binding domain to construct a bait vector. The FERRITIN sequence reported in our previous work was cloned from P. esculenta and used to amplify the coding region of the FERRITIN gene through PCR. Afterward, the amplified product was purified by using a QIA quick Gel extraction kit (Qiagen). The P. esculenta FERRITIN sequence was ligated with PGEM‐T by T4 DNA ligase. The recombinant vector was then transformed into competent E. coli DH5α and grown on an LB plate with ampicillin, X‐gal, and IPTG. The plasmid from a white colony was extracted and identified through PCR. Positive plasmids were further sequenced and compared with FERRITIN.

The PGEM‐T‐FERRITIN vector was digested at sfiI sites; after gel electrophoresis was conducted, the fragments of the same size as ferritin were purified and ligated to the digested pGB vector by T4 DNA ligase. The plasmid pGB‐FERRITIN encoding the full‐length (175 amino acids) FERRITIN gene fused in the frame of the GAL4 DNA‐binding domain was constructed as the bait vector named pGB‐FERRITIN by inserting the FERRITIN gene into the sfiI sites of the pGB vector. The constructed pGB‐FERRITIN was transformed into competent E. coli DH5α and identified by sequencing.

Screening of the P. Esculenta testis library and selecting the FERRITIN interacting clones30

Ten 2 mm colonies of the bait strain Y190 (pGB‐FERRITIN) were inoculated into 150 mL synthetic dropout nutrient medium (SD/‐Trp) overnight until optical density (OD)₆₀₀=1.2, then transferred to 2000 mLYeast Peptone Dextrose Adenine medium (YPDA) (OD₆₀₀ = 0.25), and incubated until OD₆₀₀ = 0.75. The cells were collected and washed through centrifugation at 700 × g for 5 min at room temperature and then resuspended in 12 mL of TE/LiAc (10 mM Tris‐HCl pH 8.0, 1 mM EDTA, 0.1 M Lithium acetate), which was mixed with 240 μg of the testis cDNA library and 2000 μL of the pre‐denatured carrier DNA. Afterward, 50 mL of TE/LiAc/PEG (40% PEG‐4000, 10 mM Tris‐HCl pH 8.0, 1 mM EDTA, 0.1 M Lithium actetate) was added and the mixture was incubated at 30°C. After 45 min, 4.2 mL of dimethyl sulfoxide was added and the mixture was incubated at 42°C for 20 min. The cells were then collected and added to 1000 mL YPDA medium and incubated at 30°C for 60 min. Cells were again collected and resuspended in 0.9% NaCl until 7 mL was reached. Then, 20 μL of the transformant mixture was used to titer the library, and 200 μL of the transformant mixture was plated on 15 cm SD/‐Trp/‐Leu/‐His with 30mM 3‐amino‐1, 2, 4‐triazole (3‐AT) medium. The plates were incubated at 30°C until the colonies appeared (3–21 days).

The colonies grown on SD/‐Trp/‐Leu/‐His/3‐AT medium were further selected using X‐gal selective media, in which the colonies that did not exhibit a blue color were considered negative. The positive colonies selected by X‐gal were incubated onto SD/‐Trp/‐Leu/‐His/3‐AT medium and used for plasmid extraction by the Yeast‐maker^TM plasmid isolation kit in accordance with the manufacturer's instructions. The bait and the corresponding prey plasmids were used to co‐transform the Y190 yeast strain and selected in accordance with the β‐galactosidase activity. The testis cDNA inserts were amplified from the plasmid DNA. The plasmid DNA and amplified testis cDNA inserted fragments were sequenced and analyzed by comparing with the sequences from the GenBank sequence data bank (http://www.ncbi.nlm.gov.blast).

Confirmation of positive clones through co‐immunoprecipitation (co‐IP)31

In vitro, co‐IP experiments were performed to further confirm FERRITIN/FER147 interactions. The complete open reading frames (ORF) of FERRITIN and FER147 were amplified and inserted into the PCMV‐MYC and PCMV‐HA vectors, respectively. The plasmids were transfected into 293T cells cultured in Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal bovine serum (FBS) by using the Lipofectamine 2000 (Invitrogen, USA) in accordance with the manufacturer's instructions. The cells were harvested 72 h after transfection and lysed using M‐PER mammalian protein extraction reagent (Pierce) at 4°C for 30 min. Cellular debris was removed through centrifugation at 12,000 rpm for 10 min at 4°C. Cell extracts (100 μL) were then used for the input experiment. The remaining protein was divided into two portions; one sample was combined with anti‐c‐myc tag antibody (abcam) for the co‐IP experiments, and the other sample was combined with Anti‐mouse IgG (Cell Signaling Technology, cat. 7076) for the negative control. The samples were then incubated at 4°C for 6–10 h. Protein A‐agarose beads (GIBCO, USA) were added, and the samples were incubated at 4°C overnight. The immune‐precipitated samples were washed extensively with phosphate buffered saline (PBS) and then boiled in sodium dodecyl sulfate (SDS) loading buffer. The supernatant was subjected to 12% SDS–polyacrylamide gel electrophoresis (PAGE), and the resolved protein was transferred to a nitrocellulose membrane. The membrane was immune‐blotted with mouse anti‐cmyc tag antibodies and horseradish‐peroxidase‐conjugated goat anti‐mouse IgG. The signals were detected using the ECL system (Perkin Elmer, NEL100001EA).

Recombinant fer147

Construction and expression of recombinant fer14732

The complete ORF of FER147 was amplified and inserted into a Pet‐28a vector, and the recombinant plasmid pET‐28a‐fer147 was transformed into E. coli BL21 strain. Correct insertion was confirmed by sequencing. The transformants with correctly inserted plasmid were cultured in liquid LB medium supplemented with kanamycin (50 μg/mL) at 37°C with shaking. The culture was continued and bacterial growth was monitored by measuring the cell densities at OD₆₀₀. At OD₆₀₀ between 0.4 and 0.6, IPTG (400 mmol/L) was added to a final concentration of 1 mmol/L and induce the expression of fer147 for 1–5 h. The bacteria were collected every hour and the cell lysates were obtained through centrifugation for 10 min at 6000 rpm, then resuspended in PBS and lysed through sonication. Inclusion bodies were collected by centrifugation at 12,000 rpm for 20 min, and the recombinant fer147 expression was analyzed through 15% SDS–PAGE.

Purification and refolding of the expressed recombinant fer147

The fer147 inclusion bodies were dissolved in Ni–denature–GuHCl buffer (100 mM NaH₂PO₄, 300 mM NaCl, 6 M GuHCl, pH 8.0). The Ni‐IDA column (BioVision) was equilibrated with 5–10 times the bed volume of Ni–denature–urea buffer (100 mM NaH₂PO₄, 300 mM NaCl, 8 M urea, pH 8.0), and the samples were loaded. The column was then washed with five times the bed volume of the Ni–denature–urea buffer. The proteins were eluted with 5–10 times the bed volume of elution buffer Ni–denature–250 (100 mM NaH₂PO₄, 300 mM NaCl, 250 mM imidazole, 8 M urea, pH 8.0), and the eluates were collected. The column was then rebalanced with 5–10 times the bed volume of the Ni–native–0 buffer (50 mM NaH₂PO₄, 300 mM NaCl, pH 8.0). Protein concentrations were measured through bicinchoninic acid (BCA) method, and the purity was analyzed using SDS–PAGE.

The purified inclusion body protein was sequentially dialyzed against 6 M urea buffer, 4 M urea buffer, and 2 M urea buffer for 12 h each. The next treatment was administered using refolding buffer for 24 h followed by pH 7.4 PBS for 12 h. The refolded proteins were then stored at 4°C before use.

Size exclusion chromatography33

Following affinity purification, size exclusion chromatography was performed to check the molecular mass of fer147 after renaturation. Meanwhile hferritin purchased from Sigma was used as control. The affinity‐purified protein fer147 and the control hferritin were subjected to preparative size exclusion chromatography using a superdex S200 column HiLoad 10/300 GL column (GE Healthcare Life sciences) equilibrated in buffer (40 mM Na₂HPO₄, 10 Mm NaHPO₄, 25 mM NaCl, PH 7.6) at 4°C. Protein samples (1 mL) were applied to the column at concentrations ranging between 0.5 and 2 mg/mL at a flow rate of 0.4 mL/min.

Assay of heavy‐metal‐chelating ability

Heavy metal treatment

Refolded fer147, rferritin (from our lab) and hferritin (purchased from Sigma) were diluted by PBS (PH 7.4) to the same concentrations (200 μg/mL). Protein samples were placed in dialysis bags, dialyzed against 2.5 mM FeCl₂, PbCl₂, CdCl₂, CrCl₃, CuCl₂, MnCl₂, SnCl₂, or ZnCl₂ solution for 12 h, and dialyzed against pure water for 4 h; in this procedure, water was replaced 3–4 times to remove the un‐adsorbed metal ions. The samples were kept at 4°C for later use.

SEM

Through SEM, information on aggregation structure of Fer147, rferritin, and hferritin, with and without heavy metals, was obtained in the secondary electron mode. The 10 μL protein (200 μg/mL) solutions were freeze‐dried on a clean coverslip and then fixed to the slab and coated with gold in vacuum to ensure image quality during the SEM analysis.34 Protein complexes were examined under SEM (S‐3400N, Hitachi, Japan) at 5000× magnification. Image‐Pro Plus software was used to calculate the size of each protein complex. All tests were run in triplicates.

ICP–MS

Duplicated digestion runs have been performed for all samples to detect the uptake ability of the three ferritins treated by different heavy metals. A 5% dilute nitric acid solution was added to 0.1 mL of 0.2 mg/mL ferritin solution, mixed with gentle shaking, and digested with a microwave digestion system35 (MARS, CEM, U.S.). Metal elements were then determined by ICP–MS (X Series II, Thermo Fisher Scientific, US). Standard working solutions were prepared by diluting the multi‐element standard solutions with ultra‐pure water to calibrate the instrument. These standards included a 10.0 mg/L multi‐element (CLMS‐2AN) or a tin single‐element standard solution (100 mg/L, GBW [E] 080546, Beijing Shiji Aoke, China). Reagent blanks were used to determine the background reading on the day of the analysis, and “limits of quantitation,” defined as 10 times the standard deviation of the blank runs, were calculated.36

CD analysis

The CD spectra of the proteins were highly sensitive to backbone conformation and can thus be used to estimate secondary structure content.37 Different protein secondary structures yield characteristic CD spectra. Protein solutions (1.0 mL, 0.1 mg/mL) were used to analyze CD (Chirascan, Applied Photophysics, UK). A 0.1 mm optical path length was used at room temperature, the spectra were scanned at the far UV range (190–250 nm), and the nitrogen flow rate was 5 L·min⁻¹, with water used as the reference solution. The secondary structure percentage was estimated on a JASCO J‐715 CD spectro‐polarimeter (Tokyo, Japan). Results were expressed as molar ellipticity [θ] (deg cm² dmol⁻¹) based on a mean amino acid residue weight (MRW) of the ferritins. The molar ellipticity was determined as [θ]=(θ × 100MRW)/(cl) where c is the protein concentration in milligrams per milliliter, l is the light path length in centimeters, and θ is the measured ellipticity in degrees at a wavelength λ.

Statistical analysis

Data were the means of three independent biological replicates and all statistical analyses were performed using one‐way ANOVA followed by Bonferroni's post hoc test of significance with SPSS 17.0 software. Upon which P ≤ 0.05 were interpreted as indicating statistically significant differences.

Supporting information

Supporting Information Figure S1

Supporting Information Table S1

Supporting Information Table S2

Supporting Information Table S3