The human iron exporter ferroportin. Insight into the transport mechanism by molecular modeling

Introduction

Ferroportin (Fpn), the only known mammalian iron exporter, is essential for transporting iron across the basolateral surface of enterocytes into the blood circulation system and for recycling iron in the reticuloendothelial system [1]. The synthesis of Fpn is regulated via the iron-responsive element/iron regulatory proteins system at the cellular level [2] and by hepcidin at the organism level [3, 4]. Hepcidin is a small peptide that plays a key role in the regulation of iron homeostasis by virtue of its ability to bind Fpn, and induce its internalization and subsequent degradation [3, 4]. The final effect is the inhibition of iron efflux from duodenal enterocytes, macrophages, and hepatocytes into the bloodstream [5, 6]. Hepcidin expression is probably transcriptionally regulated, although the exact mechanisms are not completely defined [4, 7, 8]. The hepcidin–Fpn axis is the principal regulator of extracellular iron homeostasis in health and disease [5].

In humans, Fpn mutations are associated with type IV hemochromatosis or “ferroportin disease.” Most mutations are “loss of function” as they affect plasma membrane localization or iron export ability, and lead to a phenotype characterized by a high serum ferritin concentration, relative plasma iron deficiency, and preferential iron retention in reticuloendothelial cells [9–12]. On the other hand, some “gain-of-function” mutations do not impair the function of Fpn but rather result in hepcidin resistance and in the anomalous retention of the active protein on the cell membrane. The pathological signatures are in this case different, with a high transferrin saturation and parenchymal iron overload [9, 10, 12–15].

The topology of Fpn has not been exactly defined; in fact, no experimental data are available up to now on Fpn three-dimensional structure. Structural bioinformatics analyses of Fpn revealed that the protein conforms to the general plan of the major facilitator superfamily (MFS) transporters [16]. This superfamily represents the largest group of secondary active membrane transporters and plays an important role in multiple physiological processes [17–19]. A canonical MFS fold comprises 12 transmembrane helices (TMs) that are organized into two folded domains, an N- and a C-terminal domain, each containing six consecutive TMs [20]. In turn, these two domains display a twofold pseudosymmetry related by an axis perpendicular to the membrane plane [20]. The rotation of one domain relative to the other allows the conformational changes necessary for the “alternate access mechanism” of transport. This is achieved through cycling from an “inward-open” to an “outward-open” conformation through an intermediated “occluded” conformation and vice versa [20].

Recently, two structural models of human Fpn have been obtained by independent groups [10, 21]. The first one, built by homology modeling based on another member of the MFS, corresponds to Fpn structure in the “occluded” state and allowed to prove the crucial role of Trp42 in both the iron transport mechanism and the binding of hepcidin [10]. Most recently, a second model of Fpn in the inward-open conformation has been obtained by ab initio modeling [21]. Analysis of this model allowed identifying a potential iron binding site, centered around residues Asp39 and Asp181, whose importance was confirmed through site-directed mutagenesis [21]. In the same study, an essential role in iron transport was also established for Asp325 and Arg466, although their involvement in the transport mechanism remained undefined [21].

In this work, we describe new structural models of Fpn obtained in the three mechanistically relevant conformations (inward open, occluded, and outward open). Fpn structure was modeled using as templates all the known structures of MFS members. Twenty-two different models have been obtained and analyzed in the context of known experimental data in order to select the best model for each conformation. The three models selected provide a prediction at the atomic level of the conformational changes occurring in Fpn during iron transport and allow proposing a possible mechanism for iron export.

Materials and methods

Results

Using the template structures listed in Table 1, 22 different molecular models of Fpn were initially generated (Supplementary Material, Figure S1). All the models display the typical fold of MFS proteins with 12 TMs spanning the membrane and the N- and C-termini located on the intracellular side. For most of the models obtained, the overall quality is fairly good, as judged by PROCHECK analysis and C-score values (see Supplementary Material, Table S1). In fact, in 18 cases out of 22, the percentage of residues in the allowed regions of the Ramachandran plot is >97% and in 16 cases out of 22 the overall G-factor calculated by PROCHECK is ≥ –0.5, the threshold for high-quality three-dimensional structures [50]. Further, in 15 cases out of 22, the C-score is greater than –2.5, the typical range for molecular models being –5 to 2 [22].

In order to select the three best models representative of the inward-open, occluded, and outward-open conformation of Fpn, the 16 models with G-factor ≥ –0.5 were further screened according to the following experimentally derived considerations:

Asp39, Asp181, and Asp325, whose mutation completely abolishes the iron binding and transport ability of Fpn, form a putative iron binding site in the inward-open and occluded conformations. In a good model, these residues should be placed at a relative distance <14 Å, as observed in the structurally characterized iron binding sites [21].

Arg466 mutation significantly reduces the iron transport ability of Fpn and has been hypothesized to act as an “electrostatic switch,” which, upon iron binding, facilitates the inward to outward transition of Fpn [21]. To establish electrostatic interactions with iron, this residue should be located at a distance <10 Å (the typical Debye length in biological systems [52]), from the putative iron binding site in the inward-open and occluded conformations.

Most members of the MFS display a conserved sequence signature [Gly]⁺¹ [Asp]⁺⁵ [Arg]⁺⁹, located at the level of the cytoplasmic loop connecting TM2 and TM3, which has been named Motif A [53]. The crystal structure of the MSF transporter YajR and mutational analyses indicated that Motif A is essential to stabilize the outward-open conformation through formation of a salt bridge between the Asp and Arg residues of the motif (see Ref. [54] and references therein). The Motif A sequence signature is conserved in Fpn and is formed by residues Gly80, Asp84, and Arg88. Thus, an additional criterion for Fpn best models selection has been the formation of a salt bridge between Asp84 and Arg88 in the outward-open conformation.

The three models that better conformed to these criteria were those built using as templates the Escherichia coli glycerol-3-phosphate transporter GlpT (Ref. [30]; PDB code 1PW4) for the inward-open conformation, the E. coli symporter proton:xylose XylE (Ref. [45]; PDB code 4GBY) for the occluded conformation, and the E. coli fucose transporter FucP (Ref. [48]; PDB code 3O7Q) for the outward-open conformation.

Figure 1 shows a schematic representation of the three models, which provide a prediction at an atomic level of the conformational changes occurring in Fpn during iron transport. In the inward-open conformation, the protein displays a wide channel accessible from the intracellular side, which closes approximately in the middle of the membrane plane at the level of residues Asp39 and Asp181. Interestingly, in this conformation, Asp325 forms a salt bridge with Arg466 with its carboxyl group facing away from the Asp39-Asp181 couple (Figure 2A). In the occluded conformation, the salt bridge is broken and Asp325 moves toward Asp39 and Asp181, forming a site characterized by a high density of negative charge (Figure 2B). Indeed, ligand binding sites detection, using the LIBRA application in conjunction with a database of structurally characterized iron binding sites (Ref. [51]; see Materials and methods for details), reveals that the substructure formed by His32, Asp39, Asp181, and Asn185 displays a stereochemistry very similar to that of the Fe(II) binding site of the metal transporter MntR [55] (Supplementary Material, Figure S2). In the outward-open model of Fpn, Asp39, Asp181, and Asp325 move away from each other, in agreement with the notion that when the protein takes up this conformation, the metal is released (Figure 2C).

Figure 1:

Schematic representation of the structural models of Fpn in the inward-open (A), occluded (B), and outward-open (C) conformations.

Figure 2:

Schematic representation of the Fpn putative iron binding site in the inward-open (top panel), occluded (center panel), and outward-open (bottom panel) conformations.

In the outward-open model, it is worthwhile to mention the presence of the canonical Motif A [Gly 80]⁺¹, [Asp 84]⁺⁵, [Arg 88]⁺⁹, in which Gly80 allows tight packing of TM2 and TM11 and Asp84 establishes electrostatic interactions with the N-terminal end of TM11 (backbone nitrogen of Ile491) and Arg88 (Figure 3). These interactions are predicted to stabilize the outward-open conformation of Fpn, as observed for other members of the MFS.

Figure 3:

Schematic representation of Motif A in the outward-open model of Fpn.

Discussion

The analysis of the Fpn models obtained in the three different functional conformations allows to rationalize the available experimental data and to formulate hypotheses on the role of specific residues in iron binding and translocation. It is worth remarking that, in the inward-open conformation, Asp325 is involved in a salt bridge with Arg466 with its carboxyl group facing away from the Asp39-Asp181 couple. Thus, in this conformation, binding of Fe(II) probably involves only the latter couple of residues. In the transition from the inward open to the occluded conformation, Asp325 moves away from Arg466 and toward the pair Asp39-Asp181. For the intrinsic methodological limits of modeling, the model has been obtained in the absence of iron and thus cannot fully represent the Fpn occluded state. However, the high negative charge density region observed in this conformation appears well fit to host an iron binding site. This hypothesis is in agreement with the experimental observation that mutations D39A, D181V, and D325A lead to significant impairment of both iron binding and iron efflux in cells transfected with the mutated Fpn [21]. Moreover, the mutation D181V has been found in members of Italian families affected by type 4 hemochromatosis [56], further confirming the crucial role of Asp181. The occluded Fpn model also supports the role of Arg466 as an electrostatic switch. In fact, while in the inward-open conformation, the arginine charge is stabilized by Asp325; in the occluded conformation, Arg466 has no charge partner and therefore is in a high energy state from an electrostatic viewpoint, being located in the partially desolvated protein interior. This high energy state could trigger the conformational change to the outward-open conformation in which, according to the corresponding Fpn model, Arg466 is fully solvent accessible. It must be considered that, in the presence of iron, the rupture of the Asp325-Arg466 salt bridge would be facilitated by electrostatic attraction of Asp325 toward the metal bound by the Asp39-Asp181 couple and electrostatic repulsion between the metal and the Arg466 guanidinium group. From this viewpoint, a very interesting observation is the stereochemical similarity detected between Fpn residues His32, Asp39, Asp181, and Asn185, and the Fe(II) binding site of the metal transporter MntR [55] (Supplementary Materials Figure S2). Interestingly, mutation of His32 in rat Fpn and mutation of Asn185 in human Fpn both lead to a nonfunctional protein [49, 57]. It appears thus likely that the site formed by His32, Asp39, Asp181, and Asn185 is the initial Fe(II) binding site whose coordination sphere would be completed by Asp325, driving the motion of TM1 toward TM7 and the transition from the inward open to the occluded state. This hypothesis is further supported by the fact that Asp39 and Asp325 are located in orthologous positions on TM1 and TM7 of two arginine residues (Arg45 on TM1 and Arg269 on TM7), which in the glycerol-3-phosphate antiporter GlpT are thought to drive the motion of TM1 toward TM7, and the transition from the inward-open to the occluded state, upon binding of the negatively charged glycerol-3-phosphate ligand [30].

Regarding the outward-open model, it is worthwhile discussing the presence of Motif A, formed by residues [Gly 80]⁺¹, [Asp 84]⁺⁵, and [Arg 88]⁺⁹, which is essential to coordinate the conformational changes of MSF transporters [54]. The presence of this motif in Fpn helps to explain the association of mutations G80S and R88G with type IV hemochromatosis [58]. In fact, the outward-open model of Fpn indicates that mutation of Gly80 would impair the correct interdomain helix packing, while mutation of Arg88 would destabilize the outward-open conformation by disruption of the salt bridge with Asp84.

In conclusion, the Fpn models described in the present work, together with available experimental data on wild-type and mutant Fpn, allow hypothesizing the following mechanism of iron transport. In the inward-open conformation, iron would bind to the Asp39-Asp181 couple, His32, and Asn185, possibly contributing to metal chelation. Iron binding would attract toward the metal coordination sphere Asp325, driving the motion of TM1 toward TM7 and the transition to the occluded state. At the same time, rupture of the Asp325-Arg466 salt bridge and repulsion of Arg466 would occur, which would generate a high energy state due to the partially desolvated arginine charge. This high energy state would be relaxed upon transition to the outward-open state, facilitated by the electrostatic interactions formed within the Motif A. At this stage, iron ligands would move apart from each other and the metal released in the extracellular space.

At present, the available data do not allow to understand if a coupling mechanism for iron transport (symport or antiport) is at play in Fpn or iron is simply transported following its concentration gradient. However, the putative transport mechanism outlined above will be useful to inspire additional mutagenesis studies that can help clarify these aspects.