Hepcidin, the hormone of iron metabolism, is bound specifically to α-2-macroglobulin in blood

The hormone hepcidin plays a role in orchestrating iron metabolism.^1-12  This peptide, originally discovered in urine as a bactericidal molecule,¹³  was later shown to be a regulator of iron metabolism.^3,5  In fact, it plays an important role in conditions of altered iron demand.^3,5 

The function of hepcidin is regulation of transmembrane iron transport.^4,5  Hepcidin binds to its cell surface receptor, ferroportin (solute carrier family 40 [iron-regulated transporter], member 1), leading to internalization and degradation of the protein complex by the lysosome.¹  Because ferroportin enables iron efflux from enterocytes, hepatocytes, and macrophages, its internalization after hepcidin binding leads to decreased iron release.¹  Hence, the hepcidin-mediated decrease in ferroportin iron export from enterocytes into blood leads to depressed intestinal iron absorption.^3,4  At the same time, iron export from hepatocytes and macrophages is blocked, which further decreases serum iron.^3-5 

Hepcidin also plays a role during inflammation, infection, and cancer.^3-5  Under these conditions, iron is shifted from the circulation into stores, making it less available.³  In anemia and hypoxia, hepcidin regulates iron availability for erythropoiesis. In the future, hepcidin may find a place in treating disease states.^3-5  Furthermore, hepcidin is predicted to become an indicator of body iron stores.

We investigated the presence of plasma hepcidin-binding molecules because the identification of such entities will provide new insights into hepcidin function. Indeed, it is known that many hormones are transported in the blood by carrier molecules,^14-16  but a specific hepcidin-binding protein has not been identified. Hence, knowledge of hepcidin transport is essential for understanding its distribution and may be important for its measurement in plasma.

In this study, we incubated blood plasma with¹²⁵I-labeled human hepcidin and then separated plasma proteins using native electrophoresis. For the identification of these molecules, the complexes were purified using a native 2-dimensional separation technique with identification by mass spectrometry.¹⁷  We identified α₂-macroglobulin (α₂-M) as the specific hepcidin-binding molecule in plasma.

Chemicals were from Sigma-Aldrich (St Louis, MO). Blood was obtained from healthy volunteers (2 females, 3 males) after the study was approved by the Ethics Committee (Institute of Hematology and Blood Transfusion, Prague) and informed consent was obtained from the participants in accordance with the Declaration of Helsinki. To obtain serum, blood was collected into vacuutainer tubes, allowed to stand 15 minutes and centrifuged (2000g for 25 minutes at 20°C). To obtain plasma, blood was collected into vacuutainer tubes and centrifuged (1500g for 5 minutes at 20°C). Plasma or serum were pooled and used for experiments. Analysis of these blood samples by the Institute of Hematology and Blood Transfusion demonstrated they were within the normal range for serum iron, serum ferritin, and transferrin saturation (Table 1^18,19 ).

Unmodified (DTHFPICIFCCGCCHRSKCGMCCKT) and M21Y-modified (DTHFPICIFCCGCCHRSKCGYCCKT) hepcidin peptides were custom-synthesized and purity was confirmed by Clonestar Peptide Services (Brno, Czech Republic). Goat anti–human α₂-M antibody (sc-8514) was from Santa Cruz Biotechnology (Santa Cruz, CA).

We used an established method for hepcidin renaturation based on dissolving hepcidin in denaturant (6 M guanidine HCl), subsequent reduction with dithiothreitol, followed by purification using reversed-phase chromatography and slow oxidation.¹ 

We compared 3 methods for determining hepcidin concentration: (1) absorbance difference at 215 to 225 nm²⁰ ; (2) quantitative determination of sulfhydryl groups²¹ ; and (3) measurement of absorbance at 205 nm.²²  All methods yielded the same concentration, and, subsequently, method 1 was used.

Hepcidin was labeled with ¹²⁵I via the protocol of Nemeth et al¹  using IodoBeads (Pierce Biotechnology, Rockford, IL). This protocol maintains the conformation of hepcidin.¹  Iodinated hepcidin was then purified using a Waters μBondapack C18 WAT027324 column (Waters, Milford, MA) prewet with methanol and equilibrated with 0.1% trifluoroacetic acid (TFA).

The biologic activity of hepcidin was determined according to Rivera.²³  Twelve to 14 days before the experiment, 5-week-old female C57Bl/6 mice (Charles River Laboratories, Wilmington, MA) were switched to a low-iron diet (Harlan-Teklad, Madison, WI). These were injected intraperitoneally with 50 μg hepcidin, controls were injected with phosphate-buffered saline (PBS). Six hours later, the mice were anesthetized with avertin and blood was collected. The blood was centrifuged at 700g for 15 minutes at 4°C and the sera collected. Serum iron and total iron-binding capacity were then analyzed (Jewish General Hospital, Montreal, Canada). Animal studies were performed with approval of the Animal Care Committee of the Lady Davis Institute for Medical Research/McGill University, Montreal, QC.

Fractionation was performed at 25°C. Blood plasma (500 μL) was diluted with 4.5 mL buffer A (50 mM Tris-HCl, pH 8.0) and loaded onto a MONO Q-5/50 GL column (GE Healthcare, Little Chalfont, United Kingdom) connected to a BioLogic HR (Bio-Rad Laboratories, Hercules, CA) fast-pressure liquid chromatography (FPLC). The fractions were eluted using a linear gradient (0-1 M) of NaCl in buffer A.

Samples were separated on a linear gradient (3%-12%) native polyacrylamide gel containing Triton X-100 in Tris-glycine buffer.¹⁷  The gel was exposed and scanned on a phosphorimager (Fuji, Cypress, CA). The radiogram was analyzed using Aida software (Raytest, Straubenhart, Germany).

Binding of hepcidin to plasma proteins or purified α₂-M was studied via size-exclusion chromatography using FPLC (BioLogic DuoFlow System; BioRad) fitted with a Superdex 200 10/300 GL (GE Healthcare) column. Samples of plasma or purified α₂-M (Sigma-Aldrich) were incubated with ¹²⁵I-hepcidin for 1 hour at 37°C, loaded into the column and eluted with 0.14 M NaCl/0.01 M Hepes (pH 7.4) at 25°C. Radioactivity in the fractions was measured with a 1480 Wallac Wizard 3′ gamma counter (Turku, Finland).

In-gel digestion and mass spectrometric analysis was performed as described.¹⁷  Proteins were reduced with 30 mM Tris-(2-carboxyethyl)phosphine hydrochloride at 65°C for 30 minutes and alkylated by 30 mM iodacetamide for 60 minutes in the dark. After overnight digestion at 37°C in buffer containing 0.01% 2-mercaptoethanol, 0.1 M 4-ethylmorpholine acetate, 10% MeCN, and sequencing grade trypsin (20 ng/μL; Promega, Madison, WI), the peptides were purified on a macrotrap column packed with polymeric reversed-phase material (Michrom BioResources, Auburn, CA). The column was connected to an LCQ^DECA ion trap mass spectrometer (Thermo Scientific, San Jose, CA) equipped with a nanoelectrospray ion source. Full scan spectra were recorded over 350 to 2000 Da followed by MS/MS scans of the 3 most intense ions in the preceding full scan. Peak lists were analyzed using SEQUEST (Thermo Fisher Scientific, Waltham, MA).

Protease-binding to α₂-M triggers a conformational change converting it to an activated form able to bind hormones.²⁴  Proteases cannot be used to activate α₂-M due to their damaging effects, which would prevent proper binding analysis. Considering this, previous workers demonstrated this conformational change can be mimicked by treatment with methylamine.^24-26  Furthermore, physicochemical and functional data suggest that methylamine activated α₂-M (α₂-M-MA) closely resembles the structure and function of protease-activated α₂-M.^25,26 

In vitro conversion of α₂-M by methylamine into the activated protein (α₂-M-MA) was achieved using an established technique where samples of α₂-M (1.4 μM; Sigma-Aldrich) were treated with 200 mM methylamine/0.05 M Tris/HCl (pH 8.1).²⁷  Unreacted methylamine was removed from α₂-M-MA via a Sephadex G-25 centrifugal column.^28,29 

α₂-M (Mw 720 000 Da) can be sedimented by ultracentrifugation.³⁰  We exploited this method to determine α₂-M-hepcidin and α₂-M-MA-hepcidin complex stoichiometry.

The mixture of α₂-M or α₂-M-MA with hepcidin (molar ratio, 1:5; at this molar ratio hepcidin is above the α₂-M and α₂-M-MA saturating concentration) was incubated at 37°C for 2 hours and then subjected to ultracentrifugation at 180 000g for 2 hours at 37°C. Control tubes contained hepcidin only. The amount of hepcidin bound to α₂-M or α₂-M-MA was calculated by subtracting the hepcidin concentration in the supernatant before and after ultracentrifugation. The number of hepcidin binding sites was calculated as the molar ratio of the bound hepcidin to total α₂-M. In preliminary experiments using the ultracentrifugation of α₂-M and α₂-M-MA without any hepcidin added, we proved by measurement of total protein (Quick Start Protein Assay Kit; Bio-Rad) that no α₂-M remained in the supernatant after ultracentrifugation, as it was completely pelleted. We also verified that the hepcidin concentration in the control tubes was the same before and after centrifugation.

To determine the equilibrium dissociation constant (K_d) for hepcidin-binding to α₂-M, samples containing a constant amount of α₂-M or α₂-M-MA and increasing concentrations of ¹²⁵I-hepcidin were incubated for 2 hours at 37°C and separated using native electrophoresis. The radiogram was quantified using Aida software (Raytest).

Under our conditions, the determination of K_d from the binding curve required correction for hepcidin depletion. Assuming that nonactivated α₂-M contains 2 binding sites for hepcidin (as indicated from ultracentrifugation experiments), we calculated the free hepcidin concentration for each well by subtraction of the bound hepcidin obtained from the y-axis values. The binding hyperbola were processed using GraphPad Prism 3.00 (GraphPad Software, San Diego, CA).

The binding data for α₂-M-MA were fitted using a multisite Hill equation.³¹  The free hepcidin concentration was calculated using the numerical solution of an implicit binding function (see equations 1FD2–3). The data analysis was based on the concerted binding reaction:

with the dissociation equilibrium constant:

where M denotes α₂-M; H, hepcidin; n, the cooperativity coefficient; and the square brackets denote concentration. We derived the modified Hill equation, that is expressed in terms of the total concentrations (hence subscript 0) of hepcidin and macroglobulin:

where Y is the ratio of the concentration of the hepcidin-α₂-M complex to that of total α₂-M. The fitting of equation 3 to the data involved iteration of the values of n and K_d combined with the numerical solution of equation 3, using NSolve in Mathematica.³²  Refinement of the choices of values for n and K_d were made by inspection of fit of the function to the data. The use of nonlinear regression was not possible because of the implicit (in Y) nature of equation 3.

J774 cells (ATCC) were grown using standard procedures³³  and used when at approximately 90% confluence to maximize ferroportin detection. Western analysis was performed by established methods^34,35  using the Invitrogen NuPAGE Novex System for optimal detection (Carlsbad, CA). Briefly, protein samples (100 μg) were separated on NuPAGE Bis-Tris 4% to 12%, 1.5 mm gels (Invitrogen) and then transferred to Invitrolon PVDF membranes. The primary antibodies used were anti-ferroportin (1/2500; Dr D. Haile, Audie Murphy Hospital, San Antonio, Texas) and anti–β-actin (1/10 000; clone AC-1; Sigma-Aldrich). Secondary antibodies used were anti–rabbit (1:2000; Sigma-Aldrich) and anti–mouse (1:10 000; Sigma-Aldrich) conjugated with horseradish peroxidase. Incubation with the primary antibody was for 2 hours at room temperature (RT), and with the secondary antibody for 1 hour at RT both in 10% skim milk (Tris-buffered saline/0.1% Tween 20; pH 7.4). The protein bands were visualized using ECL (GE Healthcare). Bands on x-ray film were quantified by scanning densitometry and analyzed using Quantity One (Bio-Rad).

Results were expressed as mean plus or minus SD. Data were compared using the Student t test. Results were considered significant for P values less than .05.

We hypothesized that similar to other hormones, hepcidin circulates bound to specific binding protein(s).^16,36  We examined the presence and identity of binding proteins using ¹²⁵I-hepcidin and native nondissociative separation techniques.

Synthetic hepcidin must be renatured to be a soluble and effective signaling molecule. This was performed using established methods and the biologic activity tested using mice.¹  Active hepcidin can be labeled by ¹²⁵I and used as a tracer. Because tyrosine, which is considered to be optimal for efficient radio-iodination, is not present in natural human hepcidin, modified hepcidin with a Met/Tyr²³  substitution (M21Y modified) was also assessed, as in previous studies.^1,6  This modified hepcidin was compared with unmodified (physiologically relevant) hepcidin.

In preliminary experiments, we compared the ¹²⁵I-labeling efficiency of both unmodified hepcidin (DTHFPICIFCCGCCHRSKCGMCCKT), containing 2 histidines and 2 phenylalanines, which can be iodinated,³⁷  and the M21Y-modified human hepcidin. We performed 2 parallel labelings of the same amount of either peptide using equal amounts of ¹²⁵I. Interestingly, both forms of hepcidin were labeled with similar specific activity (unmodified hepcidin, 1.6 μCi/μg; modified hepcidin, 2.1 μCi/μg). Hence, under these labeling conditions, tyrosine modification of hepcidin only slightly increased bound ¹²⁵I.

To avoid the possible problem that M21Y-modified hepcidin may behave differently from unmodified hepcidin, we used only unmodified (physiologically relevant), ¹²⁵I-labeled, HPLC-purified (Figure 1A) human hepcidin in all further experiments described below. Importantly, this peptide was shown to display physiologic activity by significantly (P < .01) decreasing the saturation of transferrin with iron after it was injected into mice (Figure 1B), demonstrating its utility for further studies. There was no difference in the ability of labeled or nonlabeled hepcidin to decrease transferrin iron saturation, demonstrating the functional integrity of the peptide.

After incubation of human plasma or serum with ¹²⁵I-hepcidin and subsequent separation using nondenaturing native electrophoresis, 2 major bands (labeled as “?”) were detected corresponding to complexes of ¹²⁵I-hepcidin with unknown molecules (Figure 2A). The migration of the 2 bands on Figure 2A are markedly different, at least in part, probably due to their differences in M_r. Free hepcidin is not visible, because as a cationic peptide, it migrates out of the sample well toward the cathode. For purification of these protein complexes and their identification by mass spectrometry, we required a nondenaturing separation technique that preserves ¹²⁵I-hepcidin-binding protein interactions and enables high resolution separation of the many plasma proteins. To meet these requirements, we used an established nondenaturing, 2-dimensional separation technique to identify the hepcidin-binding plasma protein(s).¹⁷  This method combines FPLC with native electrophoresis and is compatible with mass spectrometry.¹⁷ 

Fractions of plasma from the first dimension of the preparative separation using FPLC (Mono Q column; Figure 2B) were collected, and aliquots from each were trace labeled with ¹²⁵I-labeled hepcidin with which it was incubated for 1 hour at 37°C. The labeled fractions were then separated by native 3% to 12% gradient polyacrylamide gel electrophoresis (PAGE) as a second dimension of separation (Figure 2C). Fraction 13 contained only the lower, rapidly migrating band in Figure 2A, while fractions 15 to 17 contained both the lower and top bands (Figure 2C). Thus, the latter fractions were pooled and used for further analysis, described in the next section below, to identify the ¹²⁵I-hepcidin-binding molecules. It was of interest to note that the upper band in fraction 15 consisted of a doublet (Figure 2C). This could correspond to the activated and nonactivated forms of α₂-M that are known to migrate slightly differently^24,38  (see also Figure 7).

To identify the uppermost hepcidin-binding protein shown on Figure 2A, fractions 15 through 17 in Figure 2C were concentrated on centrifugal ultrafilters (Millipore Ultrafree NMW 3000) and resolved by native 3% to 12% gradient PAGE. When an aliquot of the pooled fractions was separated by native gradient PAGE and stained for protein with Coomassie blue (Figure 3A lane 1), the top band was in a similar migration position as the ¹²⁵I-hepcidin-binding molecule detected by phosphorimaging after a 1 hour incubation of ¹²⁵I-hepcidin with either human plasma (Figure 3A lane 2) or pooled fractions 15 through 17 (Figure 3A lane 3). The band resulting from the addition of ¹²⁵I-hepcidin to plasma (Figure 3A lane 2) or pooled fractions 15 through 17 (Figure 3A lane 3) comigrated with a band resulting from the incubation of ¹²⁵I-hepcidin with purified α₂-M for 1 hour at 37°C (Figure 3A lane 4). It is important to note that the uppermost band was well resolved from adjacent proteins, allowing subsequent MS analysis (described in the next paragraph).

Guided by the phosphorimager scan, the top band in Figure 3A (lane 3) was cut from the gel. This band corresponded to a complex of ¹²⁵I-hepcidin with an unknown protein from fractions 15 through 17. This segment of gel was rehydrated, digested with trypsin, and subjected to MS analysis. The unknown protein was unambiguously identified by μLC-MS/MS analysis as α₂-M (P01023; Table 2). A SEQUEST search against the human SwissProt database assigned 55 peptides to the protein covering 52% of the α₂-M sequence (Figure 3B,C; Table 2). It is notable from Table 2 that other proteins were also identified, although the abundance was far lower (number of peptides: 2-6). These comigrating molecules were also tested for ¹²⁵I-hepcidin-binding, but no significant affinity was identified (data not shown). Hence, considering the low abundance of these other proteins in comparison to α₂-M and their lack of hepcidin-binding ability, they were considered contaminants.

To further confirm that α₂-M was a specific hepcidin-binding protein, we performed α₂-M immunoprecipitation experiments on human plasma followed by stripping ligands from α₂-M with strongly acidic buffer and ultrafiltration of released peptides on a Millipore Ultrafree filter, M_r cutoff 5 kDa. We proved by mass spectrometry the presence of hepcidin in the ultrafiltrate. Other peptides, known to be bound by α₂-M, were also identified, such as osteoprotegerin.¹⁴  The detectability of hepcidin in MG immunoprecipitates from different individuals varied widely, and this was probably due to the broad intra- and interpersonal levels of hepcidin.³⁹  Moreover, α₂-M was confirmed as a major hepcidin-binding protein by other approaches described as shown in the next section.

Additional evidence that α₂-M was the hepcidin-binding protein was obtained by supershift analysis (Figure 4A,B). Human plasma was incubated with ¹²⁵I-hepcidin (2.8 μM) for 1 hour at 37°C. This sample was then divided into equal portions to which increasing amounts of anti–α₂-M antibody was added (5-20 μL; concentration: 200 μg/mL) and the samples were separated using native-gradient PAGE. Increasing amounts of anti–α₂-M antibody led to a progressive supershift of the band into the sample well (Figure 4A,B). This was consistent with the formation of a high M_r complex between the antibody and ¹²⁵I-hepcidin-α₂-M complex, which, due to its large size, remained in the sample well (Figure 4A). It is relevant to note that hepcidin is a cationic peptide and thus unbound hepcidin does not migrate into the gel. Hence, in the absence of anti–α₂-M antibody (Figure 4A lane 1), some radioactivity still remained in the sample well. Control studies using a nonspecific antibody (ie, anti–cyclin D1) did not result in a supershift (Figure 4A). Collectively, these studies demonstrated that anti–α₂-M antibody super-shifts the ¹²⁵I-hepcidin-α₂-M complex, again confirming that α₂-M is a hepcidin-binding protein.

Binding of hepcidin to α₂-M was further confirmed by size-exclusion chromatography using FPLC. As shown in the inset of Figure 5, a complex of purified α₂-M and ¹²⁵I-hepcidin comigrated with the peak of radioactivity formed in blood plasma after addition of ¹²⁵I-hepcidin. Again, this evidence supports the experiments above, confirming the role of α₂-M as a hepcidin-binding protein.

As shown in Figure 2A and C, 2 major bands were identified as hepcidin-binding proteins at the top and bottom of the gel. The studies above clearly identified the top band as α₂-M. Identification of the ¹²⁵I-hepcidin-binding protein in the lower band (Figure 2 A,C) was performed in an analogous way as described above for the top band (α₂-M). The fractions 15 through 17 from the FPLC-resolved human plasma were pooled (Figure 2C), trace labeled with ¹²⁵I-hepcidin (20 nM), and resolved by native gradient PAGE. After this band was cut from the gel and analyzed by mass spectrometry for its protein components, albumin was found to be the hepcidin-binding protein in the lower band (Table 2 band 2). Moreover, a Coomassie blue stained band of albumin (Sigma-Aldrich) shown in Figure 6A (second lane) corresponded to the most abundant protein in plasma, namely albumin (Figure 6A first lane). The lower-migrating ¹²⁵I-hepcidin-binding protein in human plasma (Figure 6B first lane) comigrated with purified albumin labeled with ¹²⁵I-hepcidin and detected by phosphorimaging (Figure 6B second lane). These experiments confirmed that albumin was the lower rapidly migrating ¹²⁵I-hepcidin–binding protein in Figure 2A and C.

Hepcidin-binding to albumin was found to be nonspecific, displaying nonsaturable kinetics in the range of up to a 1000-fold molar excess of hepcidin over albumin (Figure 6C). By quantitating radioactivity in the 2 hepcidin-containing bands (Figure 6B first lane), we found that 30% of plasma ¹²⁵I-hepcidin was bound to α₂-M, and the rest was associated with albumin. Considering this observation in relation to the fact that in humans the concentration of α₂-M (∼ 2.8-5.5 μM)^40,41  is 180 to 360 times less than albumin (≈ 1 mM),⁴²  it was clear α₂-M is the major and specific hepcidin transport molecule. Because the concentration of human hepcidin in plasma varies from 0 to 1.5 μM,³⁹  it is apparent there is an excess of α₂-M. Thus, it was important to determine the affinity of α₂-M for hepcidin to assess the specificity of the interaction that is crucial for understanding its activity.

To determine the α₂-M-hepcidin affinity constant and stoichiometry, we examined ¹²⁵I-hepcidin binding to α₂-M to assess the hepcidin concentration required to saturate a constant amount of α₂-M. We could not use the measurement of free and bound ¹²⁵I-hepcidin because, in contrast to the ¹²⁵I-hepcidin-α₂-M complex, the free unbound hepcidin is a cationic peptide¹³  that does not migrate into gels. To separate the α₂-M-¹²⁵I-hepcidin complex, we performed native electrophoresis at 37°C to provide physiologically relevant temperature, ionic strength, and pH for the determination of the hepcidin-binding constant with both α₂-M and α₂-M-MA. This is an established technique used to examine interactions of physiologic ligands with α₂-M.^43,44 

Under physiologic conditions, α₂-M is activated to bind ligands by a range of proteases.^45,46  This can be closely mimicked by treatment with methylamine (see “Materials”),²⁴  and this method is preferred to protease activation that nonspecifically damages α₂-M, perturbing the results. In fact, activation by MA is the established method of choice for examining binding of ligands to activated α₂-M.^24,27,43,44  Hence, in this investigation, we assessed the binding of hepcidin to α₂-M and its MA-activated form (α₂-M-MA; Figure 7A-C). Binding of ¹²⁵I-hepcidin to nonactivated α₂-M resulted in a curve indicating a single saturable class of noninteracting ¹²⁵I-hepcidin-binding sites (Figure 7A). Fitting of these data and Scatchard analysis led to an estimate of 2 hepcidin-binding sites per α₂-M molecule, with a K_d of 177 (± 27) nM (equation 3). Using ¹²⁵I-hepcidin, we were able to demonstrate that increasing concentrations of unlabeled hepcidin compete with ¹²⁵I-hepcidin, preventing binding of the label to α₂-M (Figure 7B). These data demonstrate that the labeled peptide was both functional and competitive at concentrations approximated by the previously measured K_d in binding with the unlabeled peptide.

In accordance with previous studies,²⁰  MA-activation caused a mobility shift in α₂-M migration on native electrophoresis, suggesting a conformational change (Figure 7C). In contrast to nonactivated α₂-M (Figure 7A), activated α₂-M-MA bound hepcidin and resulted in a sigmoidal saturation curve as a function of increasing ¹²⁵I-hepcidin concentration (Figure 7D). The shape of this curve was typical of cooperative allosteric binding⁴⁷  and its analysis using a modified Hill equation (see “Methods”) demonstrated that activation of α₂-M led to the appearance of additional binding sites. Fitting to the data gave the number of binding sites as n∼3 with a K_d for each site being approximately 0.3 μM. The estimate of n∼3 represents the lowest number of hepcidin molecules bound to the α₂-M oligomer, as the Hill equation does not account for subtle variations between affinities of the sites, thus limiting precision in this calculation.

To definitively define the stoichiometry of hepcidin-binding to α₂-M-MA, ultracentrifugation was used (see “Methods”). This method enabled separation of bound hepcidin (pelleted) and unbound hepcidin (not pelleted) and demonstrated that each molecule of α₂-M or α₂-M-MA bound 1.98 + 0.11 and 4.1 + 0.19 (3 experiments) molecules of hepcidin, respectively. This value was further substantiated by the data in Figure 7D, showing that the binding curve plateaued as the hepcidin to α₂-M molar ratio was increased to 4. Collectively, our data indicate the presence of 4 hepcidin-binding sites per α₂-M-MA molecule.

To assess the functional effect of the α₂-M-hepcidin complex, studies were initiated using commercial α₂-M (Sigma-Aldrich) and J744 cells that express ferroportin^48,49  and respond to hepcidin by reducing ferroportin expression.^50,51  Cells were incubated for 6 hours at 37°C in media without FCS containing: hepcidin (0.7 μM), α₂-M (2.8 μM), hepcidin (0.7 μM) plus α₂-M (2.8 μM), albumin (2.8 μM), or hepcidin (0.7 μM) plus albumin (2.8 μM; Figure 7E). Before the addition to cells, all solutions were incubated for 1 hour at 37°C to ensure complex formation. Hepcidin only slightly (P > .05) reduced ferroportin, while α₂-M alone or α₂-M plus hepcidin significantly (P < .05 and P < .01, respectively) decreased its expression (Figure 7E). A potential reason for α₂-M alone reducing ferroportin expression may be contamination with hepcidin present in the purified α₂-M, as found in studies examining other α₂-M ligands.¹⁴  This effect of α₂-M alone was observed in different lots of the protein from Sigma-Aldrich and also when it was purchased from a different source (Sapphire Bioscience, Sydney, Australia). In contrast to the α₂-M plus hepcidin treatment, albumin plus hepcidin did not lead to any significant decrease in ferroportin (Figure 7E). These experiments demonstrate that the α₂-M-hepcidin complex is functionally active and more effective than hepcidin alone at reducing ferroportin expression. Furthermore, it should be noted that this is not a nonspecific effect, as hepcidin added to the protein control (albumin) at the same concentration as α₂-M had no significant effect on ferroportin expression. Previous studies by Kaplan and colleagues¹  have shown that the effect of hepcidin on down-regulating ferroportin is far more complete than that demonstrated in Figure 7E. However, it is notable that Kaplan et al used HEK293 cells transfected with ferroportin, which hyperexpress this protein.¹  Hence, the comparison to J744 cells cannot be readily made as the latter express ferroportin under a physiologic control mechanism. Furthermore, the study of Kaplan et al assessed membrane ferroportin expression, while we examined total cellular ferroportin, which is less likely to demonstrate ablation after only 6 hours of incubation with α₂-M–hepcidin. Other authors using western blots have also shown, in J774 cells, that the response of ferroportin to hepcidin is less than complete.⁵⁰ 

We hypothesized that in analogy to other hormones,^52-54  hepcidin would circulate bound to specific binding proteins. We used ¹²⁵I-hepcidin to demonstrate the presence of hepcidin-binding proteins in human blood. To identify these proteins by mass spectrometry, we purified ¹²⁵I-labeled hepcidin-binding protein complexes using a native 2-dimensional technique consisting of FPLC and nondenaturing PAGE.¹⁷  The hepcidin-binding proteins were identified as α₂-M and albumin (Figures 3,Figure 4,Figure 5–6).

The identity of α₂-M as a specific, high-affinity, hepcidin-binding protein was confirmed by the following evidence: (1) purified α₂-M binds ¹²⁵I-hepcidin and this complex comigrates with the band obtained by adding ¹²⁵I-hepcidin to fractionated or whole plasma as shown in Figure 3A; (2) the hepcidin-binding protein complex in plasma was recognized and super-shifted with an anti–α₂-M antibody (Figure 4), but not by a nonspecific control antibody; (3) native size-exclusion chromatography demonstrated the ¹²⁵I-hepcidin-carrier protein complex in plasma was coeluting with ¹²⁵I-hepcidin added to purified α₂-M (Figure 5); and (4) K_d measurements demonstrated specific ¹²⁵I-hepcidin-binding to 2 (α₂-M) and 4 (α₂-M-MA) high-affinity sites on the protein, with evidence of allosteric cooperativity in α₂-M-MA (Figure 7). These studies were confirmed by ultra-centrifugation, where α₂-M or α₂-M-MA bound 2 or 4 hepcidin ligands per molecule, respectively.

The current results are significant, as it has been previously thought that due to the low M_r of hepcidin (≈ 2800 Da),⁵⁵  it would be rapidly cleared by the kidney, reducing its half-life. The specific binding of α₂-M prevents this clearance, and, hence, markedly modulates the activity of the peptide. If we consider a theoretical calculation where all α₂-M is activated (K_d: 0.3 μM) and the affinity of albumin for hepcidin is weak (assumed/approximated K_d:1 mM), then approximately 11% of hepcidin would be free. This may explain the presence of some hepcidin in urine.¹³ 

This is the first demonstration that α₂-M is a high-affinity hepcidin-binding molecule. Moreover, upon activation, there is evidence of multiple allosteric binding sites on α₂-M-MA, and this has not been described for its other ligands. Further, these observations suggest a regulatory role for α₂-M on hepcidin function. It is known that α₂-M is a multifunctional plasma protein.²⁴  The first described function of α₂-M was its “trapping” and inhibition of proteases.⁵⁶  This inhibition process triggers a conformational change in α₂-M, converting it to the activated form, which binds hormones, signaling molecules, etc.²⁴  In fact, numerous molecules bind to α₂-M and α₂-M-MA, such as transforming growth factor-β types 1 and 2,^57,58  platelet-derived growth factor,⁵⁹  nerve growth factor,⁶⁰  tumor necrosis factor-α,¹⁵  basic fibroblast growth factor,⁶¹  interleukins-1, -6, and -8,^62-64  vascular endothelial growth factor,⁶⁵  growth hormone,⁶⁶  osteoprotegerin,¹⁴  leptin,⁵²  activin, and inhibin.⁶⁷  The binding affinities (K_d) of α₂-M and α₂-M-MA for their ligands are rarely described, but for some^24,68,69  it was reported to lie within 10⁻⁷ to 10⁻⁹ M, which corresponds to the affinity of α₂-M for hepcidin (ie, α₂-M, K_d = 177 nM; α₂-M-MA, K_d ≈ 0.3 μM) observed herein.

Panyutich and Ganz³⁶  demonstrated that α₂-M-MA binds anti-microbial peptides known as defensins, which have some homology to hepcidin. However, these authors did not demonstrate cooperative allosteric binding. Considering the cooperativity found upon hepcidin-binding to α₂-M-MA (Figure 7D), it is relevant to discuss that a sigmoidal binding curve is also observed when O₂ binds with the 4 heme centers of hemoglobin, leading to efficient uptake and release.⁴⁷  Analogously, the efficient binding and release of 4 hepcidin molecules to α₂-M mediated via allosteric activation may be crucial for its function. It is notable that again, like hemoglobin, α₂-M is composed of 4 subunits,²⁴  thus alterations in interactions between subunits may lead to cooperative binding.

The functionality of the α₂-M–hepcidin complex was shown in our studies where it was more effective than hepcidin alone at reducing ferroportin-1 expression (Figure 7E). The precise mechanism inducing this effect is unknown. However, ligands bound to α₂-M undergo endocytosis by binding to the α₂-M receptor.^24,70,71  This receptor is found in many cell types,⁷²  and mediates ligand uptake and delivery to endosomes and lysosomes.^24,73,74  Therefore, considering the ability of α₂-M to target its ligands to cells,²⁴  it is conceivable that hepcidin-binding to α₂-M could influence hepcidin activity. Such alterations may include the half-life of hepcidin, its tissue transfer, and the binding of hepcidin to ferroportin. It is also possible that the α₂-M–hepcidin complex binds to other receptors. Finally, the hepcidin-MG complex could play some signaling role once it is processed by the lysosome.

In conclusion, identification of α₂-M as the hepcidin carrier will lead to deeper understanding of its role in iron metabolism. Because hepcidin has diagnostic potential, any method for determination of its blood concentration must take into account hepcidin-binding proteins. This investigation is important for understanding hepcidin function and its use in diagnostic tests.

The authors thank Mrtvá Ryba and old Sumava. We also appreciate assistance from Miroslav Kušiak and Vojtěch Drbohlav from Immunotech-Beckmann Coulter, Prague, Czech Republic. We also kindly acknowledge the assistance of the following people for their careful examination and comments on the manuscript before submission: Dr Katie Dixon, Dr Danuta Kalinowski, Ms Zaklina Kovacevic, Dr David Lovejoy, and Dr Helena Mangs.

This research was supported by grants to D.V. and J.P. from the Czech Science Foundation (GACR) No. 204/07/0830, No.204/03/H066, No. 305/09/1390, Ministry of Health (MZCR) grants MZCR/UHKT No. 023736, IGA MZCR No. NR8930-4, IGA MZCR NR8317-4, IGA No. NS 10300-3, and the Ministry of Education, Youth, and Sports of the Czech Republic (MSM) projects No. LC06044, No. LC07017, No. MSM 0021620858, No. 0021620806, and Institutional Research Concept AV0Z50200510 (IMIC). This project was also supported by a fellowship and grants from the NHMRC of Australia to D.R.R, a University of Sydney Short-Term Visiting Fellowship to D.V and D.R.R. and a University of Sydney Post-doctoral Fellowship to R.S.

Contribution: G.P., J.P., K.K., I.H., P.H., P.W.K., S.S.-L., P.P., R.S., E.B., M.L.-H.H., and Y.S.R. designed and performed experiments and wrote sections of the manuscript; and D.V. and D.R.R. designed the study, obtained grant funding, and wrote the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Des R. Richardson, Iron Metabolism and Chelation Program, Department of Pathology and Bosch Institute, Blackburn Bldg D06, University of Sydney, Sydney, New South Wales, 2006 Australia or Daniel Vyoral, Institute of Hematology and Blood Transfusion, U Nemocnice 1, Prague 2, 128 20, Czech Republic; e-mail: d.richardson@med.usyd.edu.au or Daniel.Vyoral@uhkt.cz.