Abstract
Neurologic disorders such as Alzheimer's, Parkinson's disease, and Restless Legs Syndrome involve a loss of brain iron homeostasis. Moreover, iron deficiency is the most prevalent nutritional concern worldwide with many associated cognitive and neural ramifications. Therefore, understanding the mechanisms by which iron enters the brain and how those processes are regulated addresses significant global health issues. The existing paradigm assumes that the endothelial cells (ECs) forming the blood–brain barrier (BBB) serve as a simple conduit for transport of transferrin-bound iron. This concept is a significant oversimplification, at minimum failing to account for the iron needs of the ECs. Using an in vivo model of brain iron deficiency, the Belgrade rat, we show the distribution of transferrin receptors in brain microvasculature is altered in luminal, intracellular, and abluminal membranes dependent on brain iron status. We used a cell culture model of the BBB to show the presence of factors that influence iron release in non-human primate cerebrospinal fluid and conditioned media from astrocytes; specifically apo-transferrin and hepcidin were found to increase and decrease iron release, respectively. These data have been integrated into an interactive model where BBB ECs are central in the regulation of cerebral iron metabolism.
Keywords: blood–brain barrier, mathematical modeling, neurochemistry, neurovascular unit, neurovascular coupling, receptors
Introduction
A number of neurologic disorders are associated with either too much or too little iron in the brain. Excessive brain iron accumulation is reported in Parkinson's and Alzheimer's diseases, amyotrophic lateral sclerosis, neurodegeneration with brain iron accumulation, and Huntington's disease.1 Conversely, in a 2010 WHO report, iron deficiency was identified as the most prevalent micronutrient problem. Iron deficiency is associated with significant cognitive, performance, and brain structural deficits.2,3 Although brain iron concentrations remain fairly constant, recent studies by Unger et al4 suggest that maintenance of brain iron homeostasis is a dynamic process that involves both diurnal and regional regulation. Thus, elucidation of the mechanism(s) by which the brain acquires iron and regulates brain iron transport, can provide insights into the adaptive responses involved in maintaining brain iron homeostasis and those associated with neurobiologic disease.
A widely held theory on how the brain acquires iron, the ‘transcytosis mechanism' (Figure 1, Route 1 (red) proposes that circulating transferrin-bound iron (holo-transferrin, Tf) binds to transferrin receptors (TfR) on the luminal membrane of the endothelial cells (ECs) that comprise the blood–brain barrier (BBB). The Tf–TfR complexes undergo endocytosis and traverse the ECs, where they fuse with the abluminal membrane and the Tf is released into the brain extracellular space. This perspective, which views the ECs of the BBB as a simple conduit, has several flaws. It fails to provide a mechanism for regulating brain iron uptake or to account for the substantial iron requirements of the ECs. This model also fails to account for several important aspects of cerebral iron transport: (1) the observed rate of iron transport is significantly greater than transferrin import, implying a segregation of iron and Tf within the ECs, (2) there is no mechanism to induce release of the tightly bound Tf at the abluminal membrane, (3) there is no mechanism to return the receptor to the luminal membrane,5 (4) there is no mechanism for recharging apo-transferrin after the delivery of iron to neural cells,6 and (5) there is no mechanism for regulating the expression of Tf receptors in response to intracellular iron.
Figure 1.
Model for brain iron uptake and regulation. A schematic representation of potential routes for iron delivery to the brain and sites for regulation. Route 1 (red arrows) describes the widely held ‘transcytosis' mechanism iron, which proposes that circulating transferrin-bound iron (holo-transferrin, Tf) binds to transferrin receptors (TfR) on the luminal membrane of the endothelial cells (ECs) that comprise the blood–brain barrier (BBB). The Tf/TfR complex undergoes endocytosis and Tf remains bound with iron as the endocytic vesicle traverses the ECs and subsequently fuses with the abluminal membrane at which point the Tf with iron is released into the brain extracellular space. Routes 2 (green arrows) and 3 (black arrows) also use endocytosis of the Tf/TfR complex to initiate iron uptake, however, the endocytic vesicles are delivered to a Compartment of Uncoupling of Receptor and Ligand (CURL), which is present in all cells. Here the iron Fe3+ is released from the TfR, reduced to Fe2+, and transported out of the endosome by Divalent Metal Transporter 1 (DMT1) to be available either for incorporation into iron requiring proteins or for storage with ferritin. The apo-Tf in Route 2/2A remains bound to TfR and is recycled to either the luminal or abluminal membrane where it is released and the TfR is then available for to repeat the process. Route 3 envisages a separation of Tf and TfR with holo-Tf being dispatched to abluminal membrane while TfR is returned to the luminal membrane. In contrast to Routes 1 and 3, where it is proposed that the iron as Fe3+ is release as holo-Tf, in Route 2 it is envisaged that apoTf is released via TfR on the EC and the Fe2+ is released via the transporter ferroportin. It is then reoxidized to Fe3+ with either cerruloplasmin or another ferroxidase and subsequently recombined with apoTf outside the EC. The various neural cells can then take up the reconstituted holo-Tf and any subsequently released apoTf can be recycled.
In the current study, we have considered the mechanism by which iron is transported across the enterocyte in the gut as a potential model for how iron is transported across the BBB. A key element in the regulation of the passage of iron across the enterocytes is the presence of TfRs on the blood-side membranes.7 These receptors transfer Tf-bound serum iron into the enterocyte to inform the enterocyte of the body's iron status and the release of iron is further regulated by hepcidin that is released by the liver when circulating iron levels are high. In the current study, we have investigated whether this logic also applies to the brain's ECs. We investigated the distribution of Tf receptors in the EC in an in vivo model with two goals; first to determine whether Tf receptors are present on the abluminal membrane and second, whether the intracellular distribution of Tf receptors changed in EC in response to differing brain iron status. Subsequently, we applied the in vivo findings to a cell culture model of the BBB to show the presence of a regulatory system for iron release from the BBB. The in vivo model we have used is the ‘Belgrade rat' that expresses a mutation in the iron transporter, Divalent Metal Transporter 1 (DMT1). This mutation results in loss of the ability to transport iron across the epithelial cells in the gut thus leading to a systemic anemia. The DMT1 is also responsible for the release of endosomal iron that has been released from Tf in the endosomes. In reticulocytes, this mutation prevents the release of iron from the endosome and thus 80% of the iron and transferrin recycles back to the cell surface and accounts for the hypochromia and microcytosis associated with the mutation.8,9 Similarly, this mutation results in a systemic cellular iron deficiency, which also renders the brains of the Belgrade rats iron deficient.10 Thus, they provide an ideal model in which to investigate the intracellular distribution of Tf receptors in the brain ECs to test the hypothesis that the receptor distribution is dynamic and responsive to brain iron status. By combining this information with our earlier studies8,11 and the cell culture data herein, we now propose a new model for regulation of brain iron uptake in which the ECs, far from serving as a simple conduit, are the focal point for the regulation of cerebral iron homeostasis.
Materials and methods
All animal protocols described in this study were approved by the Pennsylvania State University College of Medicine Institutional Animal Care and Use Committee and are consistent with the 2013 AVMA Guidelines.
Immuno/Electron Microscopy Protocol
Male Belgrade homozygous (B/B) and heterozygous (B/+) rats and three Sprague Dawley 6 to 7 months old were transcardially perfused with 4% paraformaldehyde (Fisher, Thermo Fisher Scientific Inc., Waltham, MA, USA, #T353-500)/0.1% glutaraldehyde (EM Sciences, Hatfield, PA, USA, 111-30-8) and postfixed overnight at 4°C in the same solution. The tissue was cryoprotected by immersion in (10% to 30%) sucrose and 16 μm sections were prepared. The sections were blocked in (5%) milk and (5%) goat serum before exposure to the mouse monoclonal antibody against the TfR (Zymed, Life Technologies, Grand Island, NY, USA, #13-6800). The secondary antibody was a goat anti-mouse Alexa Fluoronanogold Fab fragment (Nanoprobes, Yaphank, NY, USA, #7202) diluted 1:25. After exposure to the secondary antibody, the tissue was fixed with 1% glutaraldehyde in phosphate-buffered saline for 10 minutes, before exposure to silver enhancer (Nanoprobes, #2012) and then osmicated (Polysciences, Warrington, PA, USA, #223). Finally, the slides were stained with 2% uranyl acetate before embedding in resin. Sections were cut at 60 to 90 nm for non-blinded ultrastructural analysis.
Isolation of Rat Microvessels
Rat microvessels were isolated from three male homozygous and heterozygous Belgrade rats and three Sprague Dawley rats using the methodologies previously described under non-blinded conditions.11, 12, 13 Typically, neuronal and astrocytic contamination was <5% as assessed by glucose transporter distribution.12
In Vitro Experiments on Bovine Retinal Endothelial Cells
For the cell culture model of the BBB, bovine retinal endothelial cells (BRECs) of either sex were grown in MCDB-131 media (Gibco, Rockville, MD, USA) supplemented with 10% FBS (Gemini, West Sacremento, CA, USA), 10 ng/mL EGF (Invitrogen, Carlsbad, CA, USA), 0.2 mg/mL ENDO GRO (VEC Technologies, Inc., Rensselaer, NY, USA), 0.09 mg/mL heparin (Fisher Scientific, Fair Lawn, NJ, USA) and antibiotic/antimycotic (Gibco). For the experiments, BRECs were grown to confluence on porous filters (0.4 μm pore size) in transwells (Corning Life Sciences, Wilkes Barre, PA, USA) coated with fibronectin (Sigma, St Louis, MO, USA). We added serum-free and EGF-free medium to the BRECs with 138 nm hydrocortisone (Sigma) for 48 hours before the experiments to enhance formation of tight junctions.
To analyze the expression of ferroportin, BRECs were seeded in 6-well plates (1 × 106cells/well) in triplicate and treated with Hepcidin (700 nmol/L), apo-Tf (0.1 mg/mL), and holo-Tf (0.1 mg/mL) in serum-free and EGF-free MCDB131 medium for 24 hours. Then, the cells were harvested and cell lysates were solubilized in RIPA buffer (Sigma) containing a protease inhibitor cocktail (Sigma). Total protein was estimated using BCA protein assay kit (Thermo Scientific, Rockford, IL, USA) and the levels of ferroportin determined by western blot analysis with a rabbit polyclonal anti-ferroportin antibody MTP 11A and for DMT1 with NRAMP-24A antibody from Alpha diagnostics (San Antonio, TX, USA). Enhanced chemiluminescence (Perkin-Elmer, Waltham, MA, USA) was performed and the blot was scanned in the Fuji Film System. Then, the blot was stripped, and probed for β-actin (rabbit polyclonal, 1 to 5,000 dilution; Sigma).
Iron Release Experiments
59Fe-NTA complex was prepared and added to the serum-free medium in the upper chamber of the transwell and incubated overnight at 37°C. The cells were rinsed and apo-Tf (0.1 mg/mL) (Sigma), holo-Tf (0.1 mg/mL) (Sigma), hepcidin (700 nmol/L) (Peptide Institute, Inc., Osaka, Japan), apo-Ft and holo-Ft (horse spleen ferritin, Sigma) in serum-free, EGF-free MCDB131 medium was placed in the lower chamber. RITC-dextran (Sigma) was included in the media in the upper chamber to monitor the integrity of the intercellular junctions in this model. 59Fe release into the lower chamber was measured by removing aliquots every hour and counting on a single channel analyzer (Canberra, Meriden, CT, USA). RITC-dextran from the lower chamber was read at 570 nm in a fluorescence plate reader (Spectra Max Gemini, Molecular Devices, Sunnyvale, CA, USA).
Experiments on Primary Rat Astrocyte Cultures and Collection of Conditioned Media
Mixed glial cell cultures were prepared from the cortex of 1- to 2-day-old Sprague-Dawley rat brains of either sex following the method described by McCarthy and de Vellis.14 Cells were plated at a density of 1 × 106/mL in 6-well plates, 6 wells per experimental condition. The next day, the astrocytes were treated with serum-free medium, serum-free medium with deferoxamine (100 μmol/L), or serum-free medium with TMH-Ferrocene (10 μmol/L) for 24 hours. Then, medium was removed and cells were rinsed and incubated in serum-free medium for 24 hours. The conditioned media from the three conditions control, iron deficient, and iron loaded were collected and used to study the influence of astrocyte conditioned medium on iron release from BBB and to study the influence of astrocyte conditioned media on TfR expression in BRECs. Subsequently, the BRECs were harvested and lysed in RIPA buffer. The total protein was estimated using micro BCA method and immunoassay performed as described previously.1
Statistical Analysis
For all the data, one-way ANOVA was performed with post-hoc Tukey's Multiple Comparison Test to compare the means of all the groups using the GraphPad Prism 4 software (GraphPad Software, Inc. San Diego, CA, USA).
Simulation Methods
The model was developed using Berkeley Madonna X (Mach-O; v 8.3.22, Berkeley Software, Berkeley, CA, USA), which is a general-purpose differential equation solver. Simulations were run over 10,000 to 200,000 simulated minutes using the Rosenbrock (stiff) Integration Method.
Monkey Cerebrospinal Fluid Specimens
Cerebrospinal fluid (CSF) was obtained from 12 rhesus monkeys (Macaca mulatta) between 6 and 8 months of age. Samples were collected under acute ketamine sedation and postprocedural analgesics administered following an Institutional animal care and use committee-approved protocol. The monkeys were selected on the basis of well-characterized hematology indices to include six healthy iron-sufficient (two female, four male) and six iron-deficient anemic animals (two female, four male). Cerebrospinal fluid specimens (1 mL, blood free) were obtained via cervical puncture and immediately frozen in an ultracold freezer at −80°C, before being shipped to the laboratory for analysis of Fe release using the in vitro EC model of the BBB.
Results
Figure 2A illustrates western blot analysis for Tf receptors and DMT1 in the ECs derived from Sprague Dawley and heterozygous and homozygous Belgrade rats. Figures 2B and 2C illustrate the subcellular distribution of the transferrin receptors in corresponding ECs in intact brain as detected by EM-immunocytochemistry and reveal the presence of Tf receptors on the abluminal membrane of the ECs. Moreover, we observed an asymmetric transferrin receptor distribution between the luminal, the abluminal, and the intracellular membranes that was significantly modulated in the heterozygous and homozygous Belgrade rats and is quantitated in Figures 2D and 2E. If transcytosis was the mechanism for iron transport in the brain (depicted as Route 1 in Figure 1), then mutations in DMT1 should have had no effect on Tf receptor distribution. However, there was clearly a profound change in Tf receptor distribution in response to the alterations in DMT1. The observation of the presence of Tf receptors on the abluminal surface led us to a second hypothesis that signaling molecules in the CSF and extracellular fluid are able to report the brain iron status to the ECs of the BBB (Route 2 is green in Figure 1).
Figure 2.
Distribution of transferrin receptors and Divalent Metal Transporter 1 (DMT1) in brain microvessels cells isolated from Sprague Dawley (SD) rats and heterozygous (B/+) and homozygous (B/B) Belgrade rats. (A) A typical western blot of isolated microvessels from the respective rat brains (n=3), which reflects the total level of Tf receptors and DMT1 in the microvessels from the three different animal models. Note that inactive DMT1 is detected in the Belgrade animals. The presence of transferrin receptors in the microvessels was detected by EM-immunostaining in the luminal, abluminal, and intracellular membranes as indicated by an → in a representative section in (B). Typical electron micrographs obtained at the level of the substantia nigra from SD, B/+, and B/B rats, respectively, are shown in (C). The enlarged area reveals the distribution of the Tf receptors among the luminal, intracellular, and abluminal membranes with immunogold labeling. (D) Reveals the absolute distribution of the receptors among the various membranes. Analysis of number of particles in each membrane was performed on 15 vessels from each of three brains/group. The receptor levels were significantly greater in SD rats than either group of Belgrade rats. The average number of particles per microvessel is SD=134.6±8.9, B/+=77.3±5.7, B/B=96.7±6.9. SD>B/+ (P<0.0001) SD>B/B (P=0.0006) B/+ versus B/B (NSD P=0.067). The fractional distribution of total cellular Tf receptors in each membrane in is shown in (E). * denotes intracellular TfR>luminal or abluminal TfR (P≤0.0001; two-way ANOVA) comparing expression in all compartments within each animal type, ¥ denotes B/B intracellular TfR<S/D intracellular TfR (P<0.005; one-way ANOVA). § denotes abluminal TfR>luminal TfR in the B/B blood vessels (P=0.0001; two-way ANOVA). No other abluminal versus luminal differences were statistically significant. ‡ denotes B/B and B/+ abluminal TfR>S/D abluminal TfR (P<0.0006; one-way ANOVA). The relative change in the level of Tf receptors in the abluminal membranes of the B/B rats is significantly increased above both the SD and B/+ rats, which is consistent with a redistribution of receptors from the intracellular membranes to the abluminal membranes as proposed in the feedback models illustrated in Figures 1 and 6 and simulated in Figure 7.
To test these hypotheses, we used a bichamber cell culture model of the BBB.15 We obtained CSF from healthy, iron-sufficient, and anemic iron-deficient infant monkeys. The results from this experiment are shown in Figure 3A and clearly show increased iron release from ECs upon exposure to CSF from iron-deficient monkeys versus CSF from iron sufficient monkeys. It is noteworthy that the extent of iron release correlates positively with the hematological indices (Mean Corpuscular Volume and Hemoglobin) across the 12 animals tested (Figures 3B and 3C).
Figure 3.
(A) Iron release from blood–brain barrier (BBB) culture model after exposure to cerebrospinal fluid (CSF) from normal or iron-deficient (ID) monkeys. 59Fe-NTA was added to the endothelial cell monolayer in the upper chamber and incubated overnight. The medium from upper and lower chambers was removed and fresh media were returned to the upper chamber and artificial CSF or CSF from either ID or iron-sufficient (IS) monkeys was added to the basal chamber. RITC dextran was also added to the upper chamber to monitor the integrity of the tight junctions in the bovine retinal endothelial cells (BRECs). Aliquots were taken from the basal chamber at the time periods indicated. Iron release from BRECs increased significantly when CSF from ID monkeys was placed in the lower chamber compared with the iron release for control CSF. The increased iron release with IS monkey CSF compared with artificial CSF shows the presence of factors not in artificial CSF. That the iron release returns to baseline over time suggests a saturable signal consistent with our observation of Tf receptors on the abluminal membrane shown in Figure 1. (B, C) Scatterplot of individual data for CSF used in this figure, showing the consistency of the relationship between iron release from EC and the monkeys' hematological iron status. CSF was purposefully collected from six iron-deficient ones that met standard criteria for anemia (mean corpuscular volume<60 fL) and compared with specimens from six healthy iron-sufficient ones. The amount of 59Fe released at the 2-hour peak time point was inversely correlated with the monkeys' peripheral iron status, suggesting that iron transport proteins in the CSF were upregulated in an attempt to promote iron release from the ECs. Prior studies on other monkeys have shown that this compensatory response is only partially successful, however, and the brain of an anemic monkey remains iron deficient for at least 3 to 4 months after its hematology and peripheral iron status have been restored by consuming iron-rich foods or regular ferrous sulfate treatments. These data indicate that there is communication between brain iron uptake and peripheral iron status.
We subsequently investigated whether signaling to the BBB could be directed from astrocytes. As shown in Figure 4A, media from astrocytes exposed to iron chelation are associated with increased iron release as compared with media from the control astrocytes. In contrast, the media from iron-loaded astrocytes are associated with decreased iron release. Moreover, astrocyte-conditioned media also affect TfR regulation. Media from iron-deficient astrocytes increase the expression of EC TfRs whereas media from iron-loaded astrocytes decrease the expression of TfRs (Figure 4B). These data not only reveal the capacity of astrocytes to influence iron release, but also show a physiologic response within the EC. These experiments reveal that there is regulation of TfRs in ECs and the regulation is responsive to the brain's iron status and concur with our recent demonstration of iron regulatory proteins being present in the brain microvasculature.8 The return of the Fe release to baseline levels in both Figures 3 and 4 suggests the formation of a steady state between the EC and the basal chamber.
Figure 4.
Release of iron from the blood–brain barrier (BBB) is regulated by signaling proteins released by astrocytes. Rat primary astrocytes were grown in standard medium (control) or media containing either TMH-ferrocene to iron load or Desferal, an iron chelator, to iron deplete the cells. After 24 hours, the media were replaced with fresh medium after rinsing the cells thoroughly. (A) The experimental astrocyte media were allowed to equilibrate for 24 hours and then removed and placed in the lower chamber of EC cultures. The EC had been previously loaded with 59Fe. The media from the basal chamber of the EC were sampled for the presence of 59Fe over the time course indicated in the graph. Iron release from the EC significantly increased with medium from iron-chelated astrocytes in lower chamber whereas medium from iron-loaded astrocytes in the lower chamber was associated with decreased iron release. The RITC dextran levels in the lower chamber were monitored to determine the integrity of the tight junctions. These data are consistent with our model that there is iron release from the EC and that the rate of release can be altered by signaling proteins. These data show that the signaling proteins are released by astrocytes. (B) This graph shows that experimental media from astrocytes altered expression of Tf receptors in EC. Thus, indicating that media contained signals that not only affected iron release (as shown in A), but also induced changes in the EC intracellular iron status resulting in altered expression of Tf receptors in a predictable direction (increased when sensing iron deficiency and decreased when iron loaded).
We further explored the regulatory mechanisms governing iron release from ECs using specific candidate proteins found in the CSF: apo- and holo-transferrin, h-ferritin, erythropoietin, and hepcidin (Figure 5). Hepcidin acutely reduces the release of iron Fe59, whereas apo-transferrin promotes its release while holo-transferrin, ferritin, and erythropoietin are without detectable effect. The presence of an iron transporter is clearly required to account for these results and ferroportin is the obvious candidate given that it has already been shown in human and rat brain microvasculature.8,16,17 The presence of ferroportin in the BRECs was confirmed (Figure 5) and we further showed that the levels can be reduced by the presence of hepcidin and holo-transferrin, but not apo-transferrin.
Figure 5.
(A) Iron is released from endothelial cells (ECs) and the amount released can be modulated by candidate signaling proteins. To show that the stored iron could be released from the ECs, ECs were preloaded with 59Fe-NTA for 12 hours. After the 12-hour incubation period, apo-Tf (0.1 mg/mL), holo-Tf (0.1 mg/mL), hepcidin (700 nmol/L), erythropoietin (EPO 4 mU/mL), horse spleen apo-Ft (50 nmol/L), or horse spleen holo-Ft (50 nmol/L) were placed in the lower chamber. These concentrations were chosen as reflective of the levels normally found in the cerebrospinal fluid (CSF). After 1 hour, the media from the lower chamber were sampled (100 μL) and every subsequent hour for 5 hours. During this 5-hour period, dextran labeled with FITC is present in the upper chamber as a control for integrity of the tight junctions and the media from the lower chamber are monitored for the presence of FITC-dextran over the course of the experiment. The amount of 59Fe (measured as Gamma counts) released into the lower chamber over 5 -hour culture is plotted against the baseline rate of release over a 5-hour time period. There was no change in dextran content over the time period measured (data not shown). (B, C) Ferroportin is expressed in bovine retinal endothelial cells (BRECs) and is responsive to hepcidin and transferrin. Western blot analysis was performed to determine ferroportin protein expression in BRECs after 24-hour treatment with hepcidin, apo-Tf, and holo-Tf (B). Ferroportin mRNA expression relative to internal actin control (%) significantly decreased when the cells were treated with hepcidin (Hepc) and holo-Tf compared with the control BRECs (**P<0.01) (C). The expression of ferroportin significantly decreased with holo-Tf treatment compared with apo-Tf-treated BRECs (*P<0.05) and the control BRECs (**P<0.01). The bars represent mean value for each group and the error bars represent the s.e.m. for each group.
The novel findings for iron transport and regulation including redistribution of TfRs determined from the respective electron micrographs resulted in a model illustrated in Figure 6, that significantly expands the existing paradigm of receptor mediated transcytosis at the BBB. The model was generated using known values for both iron and Tf concentrations and their uptake and release kinetics in both control and Belgrade rats. The model includes estimates for iron, holo-transferrin and apo-transferrin release, holo-transferrin regeneration, and rates of neural cell transferrin/apo-transferrin recycling and CSF clearance. In establishing the model, it became necessary to invoke both an iron storage component within the EC and a mechanism(s) to curtail iron release to prevent brain iron overload. Moreover, in order for iron to be released into the brain, we were required to invoke the presence of an iron transporter on the abluminal membrane consistent with our observation for the presence of ferroportin and a mechanism to regenerate holo-transferrin in the brain interstitium. To integrate the data presented above and to formulate the conceptual model presented in Figure 1, some assumptions are necessary. The first requires that all neural cells that acquire iron via Tf will release apo-transferrin, thereby establishing a steady state between iron released by ECs and neural cell iron turnover. In this context, it is important to note that an iron excess beyond fully saturated Tf has been consistently reported in the CSF under normal conditions.18,19 In our model (Figure 1), Route 1 depicts the transcytosis model of brain iron uptake and Route 2 illustrates the pathway for iron uptake. The important distinctions between the two models are (1) Route 2 provides a mechanism to account for the iron requirements of ECs; (2) it provides a mechanism by which the iron released within ECs can become available to the brain and thus accounts for the asymmetric release of iron and transferrin described previously;18,19 (3) it proposes the presence of an iron reservoir within the EC and thus provides a mechanism through which iron regulatory proteins could regulate the expression of iron uptake and storage proteins in response to iron depletion or excess;8 (4) it provides an explanation as to why the brains of Belgrade rats are iron deficient despite Fe saturation of the circulating Tf and can also account for the increase in Tf receptor number and redistribution shown in Figure 2; (5) it provides for the regulation of iron release from the BBB by proteins generated in response to brain iron status and which act on the abluminal surface of the EC. In our model, we assume the CSF is in dynamic equilibrium with the extracellular fluid and that molecules in both compartments will turn over at a normal rate.18 More findings and details supporting this conceptual schema, including simulations of the redistribution of TfRs in the control, heterozygous, and homozygous Belgrade rats, in response to modulations in the ability of DMT1 to release iron from the endosomes of the control and Belgrade rats are simulated in Figure 7 (see also the website found in Supplementary S1). We altered the extent to which DMT1 enables release of iron from the endosome to model iron transport in homo- and heterozygous Belgrade rats. A diagrammatic version of the model in Figure 1 in which the individual rate constants for each step are identified is illustrated in Figure 6 and simulated in Figure 7. The corresponding derivations for each rate constant are described in Supplementary Tables S1 and S2. Where possible, we used published concentrations for serum and CSF Tf and iron concentrations observed in both control and B/B rats.18,20, 21, 22 We also used binding constants for TfR recycling rates derived by us,23,24 and others. We predicted concentrations of holo- and apo-transferrin under the various conditions. While interstitial Fe3+ promotes hepcidin release, which in turn inhibits ferroportin, we simulated this process (Figure 7) in a more simplified manner as hepcidin inhibition of ferroportin via a simple Michaelis-Menten inhibition in which the Ki for hepcidin inhibition of ferroportin is 0.1 μmol/L. Apo Tf stimulation of ferroportin is modeled as a simple Michaelis-Menten stimulation characterized by a stimulation factor S and a K0.5 for Apo-Tf stimulation of 0.3 μmol/L. Supplementary Table S1 illustrates the equations describing the flow of transferrin, iron, and receptors between cells and subcellular compartments; Supplementary Table S2 provides predictions of iron and transferrin flows across the BBB.
Figure 6.
The model describes transfer of Fe3+ and transferrin from serum, across the blood–brain barrier (endothelial cell, EC) to the interstitium and a ‘brain cell'. Key elements include: (1) Fe2+ export from intracellular endosome to cytoplasm (described by k9 and k19), ferroportin-mediated Fe export to interstitium (described by k11 and k20), low affinity/low capacity Fe export from EC to serum (described by J41) and hepcidin feedback inhibition of EC ferroportin when interstitial Fe2+ rises (described by Ki). This was found to be necessary to prevent overaccumulation of Fe2+. Exported Fe2+ must be reoxidized to Fe3+ by Ceruloplasmin on astrocytes before it can recombine with apotransferrin to regenerate transferrin. Intracellular Fe2+ is sequestered by ferritin (Fr). The contents of the interstitium drain at a defined rate (sink) into the cerebrospinal fluid (CSF) and subsequently out of the brain.
Figure 7.
(A) Effects of loss of Divalent Metal Transporter 1 (DMT1) on endothelial cell (EC) luminal, abluminal, and intracellular transferrin receptor content. Total transferrin receptor in each compartment includes unliganded receptor, receptor complexed with transferrin and receptor complexed with apotransferrin (e.g., in compartment 1, TR1+TRTf1+TRApo1). The simulation model was run using the flows and constants described in Supplementary Tables S1 and S2. Each run required a simulated time course of 3 × 105 seconds (3.5 days) to achieve steady state. Rate constants (k1 to k7) were carefully adjusted to simulate SD (Sprague Dawley, control) data. The activity of endosomal DMT1 (described by rate constant k9) was then adjusted to zero and relative distributions of TR recomputed. Rate constants were then adjusted (see Supplementary Table S2) until experimental data (homozygous DMT1-null condition, −/−) were successfully simulated. DMT1 (k9) was then restored to full activity, control TR distributions recalculated and rate constants were adjusted again to recapitulate control TR distributions. This process was repeated until SD and homozygous DMT1-null mice data were successfully simulated by a common set of rate constants (Supplementary Table S2). DMT1 activity (k9) was then adjusted to 50% and TR distributions computed for the DMT1 +/− condition. Open bars represent experimental data and closed bars the simulated data—all produced by a single common set of rate constants summarized in Supplementary Table S1. (B) Relative Iron (Fe) and transferrin (Tf) uptake by the brain (J13 and Tf uptake respectively in Supplementary Table S1) in control (SD), homozygous (−/−) and heterozygous (+/−) DMT1 knockouts. After simulating TR distributions in Figure 6A, iron (solid bars) and apo-transferrin (empty bars) uptake was calculated for SD (red), homozygous (green), and heterozygous (blue) DMT1 knockouts. (C) Effect of interstitial [apo-transferrin] on iron uptake by the brain. Ordinate: relative iron uptake; abscissa: interstitial [apo-transferrin] μmol/L. (D) Effect of interstitial [Fe2+] on iron uptake by the brain. Ordinate: relative iron uptake; abscissa: interstitial [Fe2+] μmol/L.
Discussion
Our model for the regulation of iron transport across the BBB (Figure 1) concurs with reports on how iron transport across intestinal enterocytes is regulated (for review, see Oates et al7 and Morgan and Oates25). Iron release from enterocytes involves extracellular hepcidin and feedback modulated by apo-transferrin/holo-transferrin ratio in the serum that reflects the iron status of the body. Herein, we have showed an analogous system in the brain with regulation of iron release and a feedback mechanism that is receptive to signals generated by different neural cells in response to their respective iron status. The first clue pointing to the existence of this feedback mechanism was the discovery of significant numbers of TfRs on the abluminal membrane of ECs, which progressively increase in the ECs of the heterozygous and homozygous Belgrade rats, respectively. These rats have a mutation in the protein DMT1, which is responsible for releasing iron from the endosomes that make up the Compartment of Uncoupling of Receptor and Ligand, in Figure 1. In the absence of functional DMT1, iron release into the cytoplasm of all the cells of the body, including ECs, is compromised. We previously reported the presence of DMT1 in ECs of the BBB both in rats26 and humans.8
Although there is not universal agreement on this point,27 the requirement for a mechanism for iron to enter the EC labile iron pool is acknowledged in the literature28 and required by our model. The studies on Belgrade rats showed that despite high circulating Tf levels, brain iron uptake was compromised.22 This outcome is clearly contrary to the predictions of the transcytosis model, but is consistent with the view that the luminal TfR is not rate-limiting for the release of iron into the brain. We used the data of Knutson et al8 and Farcich and Morgan29 in the current model, and have indicated that iron release in the EC is reduced by 50% and 100% in B/+ and B/B, respectively, providing a compelling argument that iron is released into the cytoplasm of the ECs forming the BBB and can be subsequently released into the brain. The presence of the membrane iron export protein ferroportin in EC cultures30 (Figure 5) and in vivo 31 indicates a transporter for non-transferrin mediated iron release from the EC. These observations coupled with prior reports on the existence of ferritin, the iron storage protein in ECs further strengthen the analogies between how enterocytes and ECs of the BBB regulate iron release and monitor signaling feedback.
Although our BBB model has some similarity to the gut, distinct differences exist. For example, there is no transferrin-mediated uptake of iron from the lumen of the gut, but rather the acidic environment of the gut promotes iron uptake via DMT1 directly.32 The release of iron from the gut via ferroportin is clearly analogous to the brain, although at least in cultured ECs, there also appears to be iron export via Tf.33,34 In the current model, we have accorded 0% however, the fraction of the holo-transferrin to apo-transferrin clearly remains to be determined in vivo and amended within the model. The ability of apo-transferrin to induce iron release from the ECs has not been shown in the gut models and represents a significant new insight into release of iron from cells. In enterocytes, as Fe2+ iron is exported, it is immediately oxidized to Fe3+ by hephaestin and then bound to available apo-Tf.35 In the brain, we proposed that the oxidation of the released Fe2+ via ferroportin is mediated by ceruloplasmin. Ceruloplasmin is reportedly present on the end feet of astrocytes that engulf the ECs.36,37 In our experiments in which we used CSF or astrocyte conditioned media, there was a return of Fe release to baseline suggesting the establishment of a steady state. The return of Fe to the EC would require conversion of the Fe2+ to Fe3+ and binding to apoTf. This conversion requires the presence of a soluble ferroxidase and ceruloplasmin has recently been proposed as a potential candidate by McCarthy and Kosman.38 Additional support for a role for ceruloplasmin is provided by the observation in a mouse model of aceruloplasminemia, in which ferritin expression is increased in astrocytes, suggesting that these cells have accumulated iron. The mechanism by which iron accumulates in astrocytes is not known; they do not express TfRs, but have been shown to express DMT1.26 The secretome of the astrocytes and how it is affected by their iron status is under investigation. On the basis of our model, if hepcidin is released by astrocytes, as in the liver to signal iron sufficiency, then iron accumulation by astrocytes would reduce iron release from ECs, which is indeed what we observed in Figure 4. It should be noted that the concentration of apo-transferrin in the basal chamber in our model is entirely comparable to that present in the circulation and the hepcidin level represents approximately 5 to 7X the circulating concentration and 3X the apparent dissociation constant (KD) for hepcidin..39,40 To our knowledge, the concept of ambient iron status signaling in the brain has not been investigated. Moreover, the studies using monkey CSF provide striking evidence that the systemic iron status is related to release of iron into the brain. This observation suggests a heretofore, unappreciated relationship that could represent novel insights into communication between systemic iron status and brain iron uptake.
The authors declare no conflict of interest.
Footnotes
Supplementary Information accompanies the paper on the Journal of Cerebral Blood Flow & Metabolism website (http://www.nature.com/jcbfm)
Supplementary Material
References
- Zecca L, Youdim MB, Riederer P, Connor JR, Crichton RR. Iron, brain ageing and neurodegenerative disorders. Nat Rev Neurosci. 2004;5:863–873. doi: 10.1038/nrn1537. [DOI] [PubMed] [Google Scholar]
- Lozoff B, Beard J, Connor J, Barbara F, Georgieff M, Schallert T.Long-lasting neural and behavioral effects of iron deficiency in infancy Nutr Rev 200664S34–S43., discussion S72-91. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Beard JL, Connor JR. Iron status and neural functioning. Ann Rev Nutr. 2003;23:41–58. doi: 10.1146/annurev.nutr.23.020102.075739. [DOI] [PubMed] [Google Scholar]
- Unger EL, Earley CJ, Beard JL. Diurnal cycle influences peripheral and brain iron levels in mice. J Appl Physiol (1985) 2009;106:187–193. doi: 10.1152/japplphysiol.91076.2008. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Klausner RD, Van Renswoude J, Ashwell G, Kempf C, Schechter AN, Dean A, et al. Receptor-mediated endocytosis of transferrin in K562 cells. J Biol Chem. 1983;258:4715–4724. [PubMed] [Google Scholar]
- Dickinson TK, Connor JR. Histological analysis of selected brain regions of hypotransferrinemic mice. Brain Res. 1994;635:169–178. doi: 10.1016/0006-8993(94)91436-2. [DOI] [PubMed] [Google Scholar]
- Oates PS, Thomas C, Morgan EH. Transferrin receptor activity and localisation in the rat duodenum. Pflugers Archiv: European journal of physiology. 2000;440:116–124. doi: 10.1007/s004240000256. [DOI] [PubMed] [Google Scholar]
- Knutson M, Menzies S, Connor J, Wessling-Resnick M. Developmental, regional, and cellular expression of SFT/UbcH5A and DMT1 mRNA in brain. J Neurosci Res. 2004;76:633–641. doi: 10.1002/jnr.20113. [DOI] [PubMed] [Google Scholar]
- Farcich EA, Morgan EH. Uptake of transferrin-bound and nontransferrin-bound iron by reticulocytes from the Belgrade laboratory rat: comparison with Wistar rat transferrin and reticulocytes. Am J Hematol. 1992;39:9–14. doi: 10.1002/ajh.2830390104. [DOI] [PubMed] [Google Scholar]
- Burdo JR, Martin J, Menzies SL, Dolan KG, Romano MA, Fletcher RJ, et al. Cellular distribution of iron in the brain of the Belgrade rat. Neuroscience. 1999;93:1189–1196. doi: 10.1016/s0306-4522(99)00207-9. [DOI] [PubMed] [Google Scholar]
- Burdo JR, Simpson IA, Menzies S, Beard J, Connor JR. Regulation of the profile of iron-management proteins in brain microvasculature. J Cereb Blood Flow Metab. 2004;24:67–74. doi: 10.1097/01.WCB.0000095800.98378.03. [DOI] [PubMed] [Google Scholar]
- Maher F, Vannucci SJ, Simpson IA. Glucose transporter isoforms in brain: absence of GLUT3 from the blood-brain barrier. J Cereb Blood Flow Metab. 1993;13:342–345. doi: 10.1038/jcbfm.1993.43. [DOI] [PubMed] [Google Scholar]
- Simpson IA, Appel NM, Hokari M, Oki J, Holman GD, Maher F, et al. Blood-brain barrier glucose transporter: effects of hypo- and hyperglycemia revisited. J Neurochem. 1999;72:238–247. doi: 10.1046/j.1471-4159.1999.0720238.x. [DOI] [PubMed] [Google Scholar]
- McCarthy KD, de Vellis J. Preparation of separate astroglial and oligodendroglial cell cultures from rat cerebral tissue. J Cell Biol. 1980;85:890–902. doi: 10.1083/jcb.85.3.890. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Antonetti DA, Wolpert EB, DeMaio L, Harhaj NS, Scaduto RC., Jr Hydrocortisone decreases retinal endothelial cell water and solute flux coincident with increased content and decreased phosphorylation of occludin. J Neurochem. 2002;80:667–677. doi: 10.1046/j.0022-3042.2001.00740.x. [DOI] [PubMed] [Google Scholar]
- Yang WM, Jung KJ, Lee MO, Lee YS, Lee YH, Nakagawa S, et al. Transient expression of iron transport proteins in the capillary of the developing rat brain. Cell Mol Neurobiol. 2011;31:93–99. doi: 10.1007/s10571-010-9558-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wu LJ, Leenders AG, Cooperman S, Meyron-Holtz E, Smith S, Land W, et al. Expression of the iron transporter ferroportin in synaptic vesicles and the blood-brain barrier. Brain Res. 2004;1001:108–117. doi: 10.1016/j.brainres.2003.10.066. [DOI] [PubMed] [Google Scholar]
- Bradbury MW. Transport of iron in the blood-brain-cerebrospinal fluid system. J Neurochem. 1997;69:443–454. doi: 10.1046/j.1471-4159.1997.69020443.x. [DOI] [PubMed] [Google Scholar]
- Moos T, Morgan EH. Kinetics and distribution of [59Fe-125I]transferrin injected into the ventricular system of the rat. Brain Res. 1998;790:115–128. doi: 10.1016/s0006-8993(98)00055-9. [DOI] [PubMed] [Google Scholar]
- Ueda F, Raja KB, Simpson RJ, Trowbridge IS, Bradbury MW. Rate of 59Fe uptake into brain and cerebrospinal fluid and the influence thereon of antibodies against the transferrin receptor. J Neurochem. 1993;60:106–113. doi: 10.1111/j.1471-4159.1993.tb05828.x. [DOI] [PubMed] [Google Scholar]
- Moos T, Morgan EH. The metabolism of neuronal iron and its pathogenic role in neurological disease: review. Ann NY Acad Sci. 2004;1012:14–26. doi: 10.1196/annals.1306.002. [DOI] [PubMed] [Google Scholar]
- Moos T, Morgan EH. The significance of the mutated divalent metal transporter (DMT1) on iron transport into the Belgrade rat brain. J Neurochem. 2004;88:233–245. doi: 10.1046/j.1471-4159.2003.02142.x. [DOI] [PubMed] [Google Scholar]
- Davis RJ, Faucher M, Racaniello LK, Carruthers A, Czech MP. Insulin-like growth factor I and epidermal growth factor regulate the expression of transferrin receptors at the cell surface by distinct mechanisms. J Biol Chem. 1987;262:13126–13134. [PubMed] [Google Scholar]
- Roskams AJ, Connor JR. Aluminum access to the brain: a role for transferrin and its receptor. Proc Natl Acad Sci USA. 1990;87:9024–9027. doi: 10.1073/pnas.87.22.9024. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Morgan EH, Oates PS. Mechanisms and regulation of intestinal iron absorption. Blood Cells Mol Dis. 2002;29:384–399. doi: 10.1006/bcmd.2002.0578. [DOI] [PubMed] [Google Scholar]
- Burdo JR, Menzies SL, Simpson IA, Garrick LM, Garrick MD, Dolan KG, et al. Distribution of divalent metal transporter 1 and metal transport protein 1 in the normal and Belgrade rat. J Neurosci Res. 2001;66:1198–1207. doi: 10.1002/jnr.1256. [DOI] [PubMed] [Google Scholar]
- Moos T, Skjoerringe T, Gosk S, Morgan EH. Brain capillary endothelial cells mediate iron transport into the brain by segregating iron from transferrin without the involvement of divalent metal transporter 1. J Neurochem. 2006;98:1946–1958. doi: 10.1111/j.1471-4159.2006.04023.x. [DOI] [PubMed] [Google Scholar]
- Moos T, Rosengren Nielsen T, Skjorringe T, Morgan EH. Iron trafficking inside the brain. J Neurochem. 2007;103:1730–1740. doi: 10.1111/j.1471-4159.2007.04976.x. [DOI] [PubMed] [Google Scholar]
- Farcich EA, Morgan EH. Diminished iron acquisition by cells and tissues of Belgrade laboratory rats. Am J Physiol. 1992;262:R220–R224. doi: 10.1152/ajpregu.1992.262.2.R220. [DOI] [PubMed] [Google Scholar]
- Hadziahmetovic M, Song Y, Ponnuru P, Iacovelli J, Hunter A, Haddad N, et al. Age-dependent retinal iron accumulation and degeneration in hepcidin knockout mice. Invest Ophthalmol Vis Sci. 2011;52:109–118. doi: 10.1167/iovs.10-6113. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Connor JR, Ponnuru P, Wang XS, Patton SM, Allen RP, Earley CJ. Profile of altered brain iron acquisition in restless legs syndrome. Brain. 2011;134:959–968. doi: 10.1093/brain/awr012. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Oates PS, Thomas C. Augmented internalisation of ferroportin to late endosomes impairs iron uptake by enterocyte-like IEC-6 cells. Pflugers Archiv: European journal of physiology. 2005;450:317–325. doi: 10.1007/s00424-005-1421-7. [DOI] [PubMed] [Google Scholar]
- Hudson DM, Krisinger MJ, Griffiths TA, Macgillivray RT. Neither human hephaestin nor ceruloplasmin forms a stable complex with transferrin. J Cell Biochem. 2008;103:1849–1855. doi: 10.1002/jcb.21566. [DOI] [PubMed] [Google Scholar]
- Burdo JR, Antonetti DA, Wolpert EB, Connor JR. Mechanisms and regulation of transferrin and iron transport in a model blood-brain barrier system. Neuroscience. 2003;121:883–890. doi: 10.1016/s0306-4522(03)00590-6. [DOI] [PubMed] [Google Scholar]
- Chen H, Attieh ZK, Dang T, Huang G, van der Hee RM, Vulpe C. Decreased hephaestin expression and activity leads to decreased iron efflux from differentiated Caco2 cells. J Cell Biochem. 2009;107:803–808. doi: 10.1002/jcb.22178. [DOI] [PubMed] [Google Scholar]
- Unger EL, Earley CJ, Thomsen LL, Jones BC, Allen RP. Effects of IV iron isomaltoside-1000 treatment on regional brain iron status in an iron-deficient animal. Neuroscience. 2013;246:179–185. doi: 10.1016/j.neuroscience.2013.04.049. [DOI] [PubMed] [Google Scholar]
- Chirasani SR, Markovic DS, Synowitz M, Eichler SA, Wisniewski P, Kaminska B, et al. Transferrin-receptor-mediated iron accumulation controls proliferation and glutamate release in glioma cells. J Mol Med (Berl) 2009;87:153–167. doi: 10.1007/s00109-008-0414-3. [DOI] [PubMed] [Google Scholar]
- McCarthy RC, Kosman DJ. Ferroportin and exocytoplasmic ferroxidase activity are required for brain microvascular endothelial cell iron efflux. J Biol Chem. 2013;288:17932–17940. doi: 10.1074/jbc.M113.455428. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Zhang DL, Senecal T, Ghosh MC, Ollivierre-Wilson H, Tu T, Rouault TA. Hepcidin regulates ferroportin expression and intracellular iron homeostasis of erythroblasts. Blood. 2011;118:2868–2877. doi: 10.1182/blood-2011-01-330241. [DOI] [PMC free article] [PubMed] [Google Scholar]
- De Domenico I, Nemeth E, Nelson JM, Phillips JD, Ajioka RS, Kay MS, et al. The hepcidin-binding site on ferroportin is evolutionarily conserved. Cell Metab. 2008;8:146–156. doi: 10.1016/j.cmet.2008.07.002. [DOI] [PMC free article] [PubMed] [Google Scholar] [Retracted]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.