Hepcidin Deficiency Protects Against Atherosclerosis

Abstract

Objective:

Inflammatory stimuli enhance the progression of atherosclerotic disease. Inflammation also increases the expression of hepcidin, a hormonal regulator of iron homeostasis, which decreases intestinal iron absorption, reduces serum iron levels and traps iron within macrophages. The role of macrophage iron in the development of atherosclerosis remains incompletely understood. The objective of this study was to investigate the effects of hepcidin deficiency and decreased macrophage iron on the development of atherosclerosis.

Approach and Results:

Hepcidin- and LDL receptor-deficient (Hamp^−/−/ Ldlr^−/−) mice and Hamp^+/+/ Ldlr^−/− control mice were fed a high-fat diet for 21 weeks. Compared to control mice, Hamp^−/−/ Ldlr^−/− mice had decreased aortic macrophage activity and atherosclerosis. Because hepcidin deficiency is associated with both increased serum iron and decreased macrophage iron, the possibility that increased serum iron was responsible for decreased atherosclerosis in Hamp^−/−/ Ldlr^−/− mice was considered. Hamp^+/+/ Ldlr^−/− mice were treated with iron dextran so as to produce a two-fold increase in serum iron. Increased serum iron did not decrease atherosclerosis in Hamp^+/+/ Ldlr^−/− mice. Aortic macrophages from Hamp^−/−/ Ldlr^−/− mice had less labile free iron and exhibited a reduced pro-inflammatory (M1) phenotype compared to macrophages from Hamp^+/+/ Ldlr^−/− mice. THP1 human macrophages treated with an iron chelator were used to model hepcidin deficiency in vitro. Treatment with an iron chelator reduced LPS-induced M1 phenotypic expression and decreased uptake of oxidized LDL.

Conclusions:

In summary, in a hyperlipidemic mouse model, hepcidin deficiency was associated with decreased macrophage iron, a reduced aortic macrophage inflammatory phenotype and protection from atherosclerosis. The results indicate that decreasing hepcidin activity, with the resulting decrease in macrophage iron, may prove to be a novel strategy for the treatment of atherosclerosis.

Introduction

Atherosclerosis is the leading cause of morbidity and mortality worldwide.¹ In early stages of atherogenesis, the vascular intimal layer accumulates oxidized low density lipoprotein (LDL) particles, which activate the overlying endothelium.² Monocytes adhere to the activated endothelium and migrate into the intima, where they differentiate into macrophages and accumulate lipids, thereby becoming foam cells.³ Although macrophages are a key cell type in the development of atherosclerotic plaques and vascular calcification,^4,5 the precise contribution of macrophage iron levels to the pathogenesis of atherosclerosis is uncertain. Previous epidemiological studies produced contradictory results: Increased serum ferritin, which is a marker for increased intracellular iron, was associated with the progression of atherosclerosis.^6-9 In contrast, a Mendelian randomization analysis showed that higher levels of serum iron, as indicated by an increased transferrin saturation or increased ferritin, were associated with a reduced risk of coronary artery disease.¹⁰

Murine models of atherosclerosis also produced conflicting results. Reduced macrophage iron, achieved through either an iron-deficient diet¹¹ or treatment with an iron chelator,¹² was associated with decreased atherosclerosis. However, a high-iron diet was also found to decrease atherogenesis.¹³ Contrary to both of these findings, Kautz and colleagues reported that iron loading of macrophages had no effect on atherosclerosis.¹⁴ Thus, in both humans and mouse models, the role of iron in the development of atherosclerosis remains to be established.

Iron is an essential element that is required for many biological processes including oxygen transport, oxidative phosphorylation and DNA repair. Increased levels of iron may, however, have adverse consequences, resulting in part from the formation of reactive oxygen species that induce cellular damage. The level of serum iron is determined by absorption of iron from the gastrointestinal tract, recycling of erythrocyte iron by macrophages and release of iron from stores in the liver.^15,16 Iron is exported from within cells by ferroportin, the only known cellular channel to transport iron out of cells. Hepcidin, a hormone synthesized predominantly by the liver, is considered the master regulator of iron homeostasis.¹⁷ Hepcidin binds to and induces degradation of ferroportin, thus decreasing serum iron levels by trapping iron within duodenocytes, hepatocytes and macrophages.

The objective of this study was to investigate the roles of hepcidin deficiency and decreased macrophage iron in the development of atherosclerosis. We combined a murine model of hepcidin deficiency with the Ldlr^−/− mouse model of atherosclerosis. We show that hepcidin deficiency, with resulting depletion of macrophage iron, significantly reduces atherosclerosis independent of its effect on serum iron.

Materials and Methods

The data that support the findings of this study are mostly available within the article and supplementary files; any additional data are available from the corresponding author upon request. The experiments performed adhere to the American Heart Association guidelines for animal atherosclerosis studies¹⁸ as well as the guidelines for consideration of sex differences in arterial pathology studies.¹⁹

Chemicals and Reagents

Near-infrared fluorescent imaging agent ProSense 750 (NEV10001EX) was purchased from PerkinElmer (Waltham, MA). Iron dextran was obtained from Watson Pharma, Inc. (Morristown, NJ). Dextran was purchased from Sigma-Aldrich (Saint Louis, MO).

Animals

All experiments involving mice were approved by the Partners Subcommittee on Research Animal Care (Protocol #2011N000133). Ldlr^−/− mice on a C57BL/6 background were purchased from Jackson Laboratories (Bar Harbor, ME). Hamp^−/− mice, originally generated by Vaulont and colleagues ²⁰ and backcrossed on a C57BL/6 background²¹ were kindly provided by Dr. Tomas Ganz (University of California, Los Angeles) and were mated with Ldlr^−/− mice to generate Hamp^+/−/ Ldlr^+/− mice. Subsequent mating produced Hamp^−/−/ Ldlr^−/− mice.

All mice were initially fed a standard chow diet (RMH 3000, LabDiets). To induce atherosclerosis, six-week-old male Hamp^−/−/ Ldlr^−/− mice and Hamp^+/+/ Ldlr^−/− controls were placed on a western-style diet (40% fat, 0.21% cholesterol, 20% casein; D12079B, Research Diets, New Brunswick, NJ) for 21 weeks. Tangirala et al observed that female Ldlr^−/− mice eating a western-style diet developed less aortic atherosclerosis than male Ldlr^−/− mice.²² Similarly, in pilot experiments, we found that female Ldlr^−/− mice developed ~50% less aortic atherosclerosis than male Ldlr^−/− mice (n=5 in each group, p=0.02). In these studies, we therefore focused on male Ldlr^−/− mice.

To assess the effect of iron overload on the development of atherosclerosis, six-week-old Hamp^+/+/ Ldlr^−/− male mice were placed on the high-fat diet for 21 weeks and received six intraperitoneal injections of either iron dextran (INFeD, Watson) or dextran alone (Sigma-Aldrich, St. Louis, MO). Mice that were injected with iron dextran received 8 mg per dose, for a total of 48 mg over the 21-week period. Control mice were injected with an equal amount of dextran by weight.

Mice were anesthetized with intraperitoneal ketamine (120 mg/kg) and xylazine (4 mg/kg). Whole blood was collected via cardiac puncture and allowed to clot. Serum was separated from red blood cells by centrifugation. The liver, spleen, and aorta were harvested from all mice. Liver and spleen were divided in half. One half was flash-frozen in liquid nitrogen for subsequent RNA isolation; the other half was placed in 10% formalin for paraffin embedding. The aorta was initially imaged (see below) and then the thoracic region was either placed in 10% formalin for subsequent Oil Red O staining or was flash frozen for RNA isolation.

Peritoneal macrophages were harvested as previously described.²³ In brief, eight-week-old Hamp^−/−/ Ldlr^−/− mice and Hamp^+/+/ Ldlr^−/− control mice were injected intraperitoneally with 1 ml of 3% Brewer’s thioglycollate medium. After four days, peritoneal macrophages were harvested by intraperitoneal lavage with 10 ml DMEM. Cells were lysed in RIPA buffer containing protease and phosphatase inhibitors (Sigma).

Near-Infrared Imaging and Quantification of Atherosclerotic Lesions

ProSense 750 (ProSense) is a near-infrared fluorescent cathepsin-activated imaging probe, which was previously shown to identify regions of increased macrophage activity.^24,25 Animals were injected with ProSense (2 μL per gram bodyweight) via the tail vein 24 hours prior to euthanasia. The aorta was dissected, separated from myocardial tissue, and imaged ex vivo using an Odyssey Imaging System (LI-COR Biotechnology, software version 3.0.16, Lincoln, NE).

Quantitative Real-Time PCR

Total cellular RNA from mouse liver, spleen, and aorta was extracted using Trizol (Invitrogen, Carlsbad, CA). To obtain cDNA, reverse transcription was performed using MultiScribe Reverse Transcriptase (Applied Biosystems, Foster City, CA), a recombinant mouse Maloney leukemia virus reverse transcriptase. TaqMan Fast Advance Master Mix (Applied Biosystems) or SYBR Select Master Mix (Applied Biosystems) was used to perform quantitative PCR. TaqMan assays were used for 18S ribosomal RNA and hepcidin mRNA. Forward and reverse primer sequences for CD68 were 5’-CGATGCCCTGCCAATCGAGATGCTGG-3’ and 5’-CCCGGGGAGCATGTCAAGGTCAAAATCG-3’, respectively. Real-time amplification and quantification of transcripts was performed using a Mastercycler Realplex (Eppendorf, Hamburg, Germany). The expression of target genes was determined using the relative C_T method and values were normalized to levels of 18S ribosomal RNA.

Immunoblot

Samples were prepared in Laemmli buffer (1x) and were separated by SDS-PAGE, transferred to polyvinylidene difluoride membranes, and probed with antibodies specific for inducible nitric oxide synthase (iNOS, Novus, NB300-605,2 μg/mL), arginase1 (Arg1, Santa Cruz, sc-271430, 0.4 μg/mL), CD11b (Abcam, ab133357, 0.06 μg/mL) or glyceraldehyde 3-phosphate dehydrogenase (GAPDH, Cell Signaling, #5174, 4.2 ng/mL). Blots were incubated with fluorescent dye-labeled anti-rabbit IgG antiserum (IRDye 800CW, LI-COR, 1:10,000) or anti-mouse IgG antiserum (IRDye 680RD, LI-COR, 1:10,000) and quantified using a LI-COR Odyssey detection system (LI-COR, Lincoln, NE).

Histology and Immunohistochemistry

After paraffin embedding, 6-μm sections were prepared from liver and spleen and were stained with Prussian blue and counterstained with nuclear fast red. For visualization of atheroma, the thoracic region of the aorta was treated with 0.2% Oil Red O solution in methanol for thirty minutes. The aorta was then exposed to a series of 78% methanol washes, followed by a xylene rinse, as previously described. ²⁶ Oil Red O staining was quantified on the basis of plaque area and intensity of staining was determined using ImageJ Software.

For immunohistochemistry, after deparaffinization and antigen retrieval, which consisted of boiling the slides in sodium citrate buffer (pH = 6.0, 10 mM) for 5 minutes, 6-μm sections of ascending aortas were stained for MAC2/galectin-3 (Cedarlane, CLCR2A00, 1:100). Sections were probed with a biotinylated secondary antibody using the Vectastain Elite ABC kit (Vector Laboratories, Burlingame, CA, USA) and counterstained with hematoxylin.

Serum Analysis

Serum iron levels were measured using the Iron-SL kit (Sekisui Diagnostics, Lexington, MA), which takes advantage of a reaction between Ferene, an iron chelating agent, and ferrous iron, forming a blue chromophore that absorbs light at 595 nm. LDL levels were determined using an AU680 Chemistry System (Beckman Coulter, Brea, CA).

Flow Cytometry of Dissociated Aortic Cells

Flow cytometric evaluation of aortic leukocytes was performed, as previously described.^27,28 Briefly, aortas were harvested from Hamp^+/+/Ldlr^−/− and Hamp^−/−/Ldlr^−/− mice and flushed with heparinized saline to minimize contamination from circulating WBCs. The aortas were enzymatically digested in a buffer containing Collagenase I, Collagenase XI, Hyaluronidase I and DNAse I. A single-cell suspension was created by passing the cells through a 70 μM filter. The cells were then washed and stained with a mixture of conjugated antibodies directed against CD45 (APC), CD206 (PE), and F4/80 (PE Cy7), all from eBioscience/ThermoFisher Scientific, and CD11b (APC Cy7) and CD38 (PerCpCy5.5) from BD Pharmingen. Cells were also stained with a viability dye (Sytox Blue, Invitrogen/ThermoFisher Scientific) and Calcein-AM (0.125 μM, Life Technologies/ThermoFisher Scientific); the intensity of Calcein-AM varies inversely with intracellular iron concentration.²⁹ Appropriate controls, including isotype control antibodies, single-stained cells and Fluorescence Minus One-controls were used. Flow cytometry was performed on a FACS Aria III machine (BD Biosciences), and data were analyzed using FlowJo software (FlowJO, LLC). The gating strategy employed is detailed in Supplemental Figure IV.

Unsupervised High-Dimensional Analysis of Aortic Leukocytes

Data from the flow cytometry experiment described above were analyzed in an unsupervised fashion using a validated clustering algorithm called FlowSOM,³⁰ which uses self-organized maps to define populations of cells based on the expression of all markers on all cells. These populations were subsequently mapped onto a dimensionality reduction algorithm.³¹ The combination of the two algorithms defines populations of cells that are similar to each other and this approach has the capacity to define unique populations of cells that may be missed by traditional manual analysis.³² The analyses were performed using the FlowSOM and t-SNE plug-ins in FlowJo software (FlowJO, LLC).

THP1 cells

For flow cytometry, THP1 cells (ATCC) in 6-well tissue culture plates were incubated with phorbol-12-myristate-13-acetate (PMA, 4 μM) for 24h in complete medium (RPMI with 10% Fetal Calf Serum) to allow the cells to become adherent to the wells. The cells were then washed with PBS and incubated in starvation medium (RPMI with 0.3% bovine serum albumin) overnight. The adherent, starved THP1 cells were incubated with ferric ammonium citrate (FAC, elemental Iron 200 μM) or FAC + Deferiprone (DFP, 100 μM) for one hour, after which the cells were treated with Dil-Labeled oxLDL (Kalen Biomedical, 10 μg/ml) for 3 hours. Some wells were incubated with an excess of unlabeled oxLDL prior to the addition of Dil-labeled oxLDL. The cells were then washed, mechanically detached and incubated with Calcein-AM (0.125 μM) and a viability dye (Sytox Blue, ThermoFisher Scientific) before flow cytometric assessment.

Statistical Analysis

Statistical analyses were performed using either GraphPad Prism 7.0 (GraphPad Software, La Jolla, CA) or Stata 13.0 (StataCorp, College Station, TX). The Shapiro-Wilk test was performed to test for normality. All variables were found to be normally distributed except for the LDL values reported in Figure 1G. Two-group comparisons were made using the Student’s unpaired t test with equal variances or the unpaired t test with Welch’s correction for unequal variances (for normally distributed variables; variances were compared using the F test) or the Mann Whitney test (for the non-normal variable in Figure 1G). Comparisons of more than two groups (Figures 3F and 3I) were made using 1-way ANOVA with the Bonferroni-adjusted post-hoc testing for multiple comparisons, because these variables were normally distributed and because variances were found to be equal by the F test. Multivariable linear regression analysis was performed to determine whether hepcidin deficiency was associated with a change in the development of aortic macrophage accumulation after adjusting for serum iron, serum LDL, and body mass. Data are reported as mean ± SEM, unless otherwise specified. A p-value ≤ 0.05 (adjusted for multiple comparisons) was considered statistically significant.

Figure 1:

Hepcidin deficiency in Ldlr^−/− mice was associated with decreased atherosclerosis. (A) Macrophage activity was decreased in the aortas of Hamp^−/−/ Ldlr^−/− mice compared to the aortas of Hamp^+/+/ Ldlr^−/− mice as determined using ProSense near-infrared fluorescence. Results from three representative pairs of mice are shown. (B) Quantification of ProSense signal from Hamp^−/−/ Ldlr^−/− mice revealed a 45% reduction in macrophage activity compared to Hamp^+/+/ Ldlr^−/− mice (p<0.0001, n=23 and 27, respectively). (C) There was decreased lipid accumulation in the aortas of Hamp^−/−/ Ldlr^−/− mice compared to the aortas of Hamp^+/+/ Ldlr^−/− mice as determined using Oil Red O stain. Aortas from the same three pairs of mice as in A are shown. (D) Quantification of the area and intensity of Oil Red O stain revealed a 74% decrease in aortic lipid in Hamp^−/−/ Ldlr^−/− mice compared to Hamp⁺/ Ldlr^−/− mice (p=0.005, n=6 for each). (E) Consistent with the observed decrease in aortic macrophage activity, there was a 67% decrease in the mRNA levels of CD68 in the aortas of Hamp^−/−/ Ldlr^−/− mice compared to controls (p=0.002, n=12 and 16, respectively). The final weight (F) and fasting serum LDL (G) were decreased in Hamp^−/−/ Ldlr^−/− mice compared to control mice (p=0.01 and 0.003, respectively). In (F) and (G), n=23 for Hamp^−/−/ Ldlr^−/− mice and n=27 for Hamp^+/+/ Ldlr^−/− mice.

Figure 3.

Aortic and peritoneal macrophages from hepcidin-deficient mice as well as human macrophages treated with an iron chelator exhibit a reduced M1 phenotype and reduced uptake of oxidized LDL. (A) Pooled aortic CD11b-positive leukocytes from Hamp^−/−/ Ldlr^−/− mice have higher median calcein fluorescence (and therefore a lower labile intracellular iron concentration) than CD11b-positive leukocytes from Hamp^+/+/ Ldlr^−/− mice. (B) CD11b-positive macrophages isolated from the aortas of Hamp^−/−/ Ldlr^−/− mice fed a high-fat diet (n=6) exhibited lower rates of CD38 positivity (an M1 marker) on flow cytometric analysis and (C) similar rates of CD206 positivity (an M2 marker) compared to CD11b-positive macrophages isolated from the aortas of Hamp^+/+/ Ldlr^−/− mice (n=5). (D) Peritoneal macrophages isolated from Hamp^−/−/ Ldlr^−/− mice 4 days after intraperitoneal injection of thioglycollate had lower protein levels of iNOS (an M1 marker) relative to arginase-1 (an M2 marker) compared to peritoneal macrophages isolated from Hamp^+/+/ Ldlr^−/− mice. No difference in CD11b protein levels was observed. (E) Compared to peritoneal macrophages from Hamp^+/+/ Ldlr^−/− mice, macrophages from Hamp^−/−/ Ldlr^−/− mice had a decreased ratio of iNOS to Arg1. (F) THP1 cells (a human monocyte cell line) treated with PMA for 24 hours were incubated in the presence or absence of LPS or DFP (an iron chelator). LPS-treated cells exhibited increased iNOS mRNA levels; however, iNOS mRNA levels were not increased in cells treated with LPS and DFP (n=5 in each group). (G) THP1 cells incubated in culture media containing 100 μM ferric ammonium citrate and labeled oxidized LDL (oxLDL) were treated with either DFP (100 μM) or vehicle. Treatment with DFP resulted in reduced oxLDL uptake by macrophages as determined by flow cytometry. A representative pair of histograms is shown. (H) Comparison of mean fluorescence intensity (MFI) after uptake of labeled oxLDL by Fe-treated or Fe and DFP-treated THP1 cells (n=6 experiments each). (I) Treatment of THP1 cells with oxLDL increased ABCA1 mRNA levels, but this increase was blunted by treatment with ferric ammonium citrate.

Results

Hepcidin Deficiency Reduces Atherosclerosis

Hepcidin is a 25 amino acid hormone that binds to and induces degradation of ferroportin, the channel that exports intracellular iron. In the absence of hepcidin, the persistence of ferroportin in the cell membrane of macrophages results in export and subsequent depletion of iron from macrophages. To investigate the effect of decreased macrophage iron on the development of atherosclerosis, adult Hamp^−/−/ Ldlr^−/− mice and Hamp^+/+/ Ldlr^−/− (control) mice were fed a high-fat diet for 21 weeks. Consistent with the phenotype of hepcidin-deficient mice,²⁰ Hamp^−/−/ Ldlr^−/− mice had a higher serum iron level than control mice (333±25 vs. 126±9 μg/dL, p<0.0001, Supplemental Figure IA). Using Prussian blue staining of histological sections, we observed that Hamp^−/−/ Ldlr^−/− mice had increased iron stores in the liver (Supplemental Figure IB) and that no iron was detected in splenic macrophages (Supplemental Figure IC). Despite the lack of detectible iron stores in the spleen of Hamp^−/−/ Ldlr^−/− mice, there was robust ferroportin expression (Supplemental Figure II) reflecting the lack of hepcidin.

Hepcidin deficiency was associated with decreased aortic macrophage activity, as indicated by a 45% reduction in ProSense signal in the aortas of Hamp^−/−/ Ldlr^−/− mice compared to Hamp^+/+/ Ldlr^−/− mice (p<0.0001, Figure 1A and 1B). Hamp^−/−/ Ldlr^−/− mice had 74% less aortic lipid accumulation than Ldlr^−/− control mice, as determined by Oil Red O staining. (p=0.005, Figure 1C and 1D). Consistent with the observed decrease in aortic macrophage activity, there was reduced MAC-2 staining (Supplemental Figure III) in the aortas of Hamp^−/−/ Ldlr^−/− mice compared to controls. A decrease in the mRNA levels of CD68 (a marker of macrophages) in the aortas of Hamp^−/−/ Ldlr^−/− mice compared to controls (67% reduction, p=0.002, Figure 1E) was also observed. Taken together, these results show that hepcidin deficiency in a mouse model of atherosclerosis is associated with increased serum iron, decreased macrophage iron, decreased aortic macrophage activity, and decreased aortic lipid deposition.

After 21 weeks on a high-fat diet, the average body weight of Hamp^−/−/ Ldlr^−/− mice was less than that of control Ldlr^−/− mice (40.4±0.9 vs 43.8±0.9 grams, p=0.01; Figure 1F). The fasting serum LDL levels were lower in Hamp^−/−/ Ldlr^−/− mice compared to Hamp^+/+/ Ldlr^−/− mice (262±39 vs 392±38 mg/dL, p=0.003; Figure 1G). There was no significant difference in serum HDL or glucose between the two groups (not shown).

Multivariable generalized linear regression models were used to investigate the contributions of serum iron, body weight and LDL levels to the protective effect of hepcidin deficiency against atherosclerosis. In a bivariable model, the independent effects of hepcidin deficiency and serum iron levels on ProSense signal were determined. Whereas hepcidin deficiency was associated with a 66% reduction in aortic macrophage activity (p<0.001, Table 1), serum iron levels were associated with increased aortic macrophage activity (5% increase for every 50 ug/dL increase in iron, p=0.017). In multivariable analysis, the association between hepcidin deficiency and decreased atherosclerosis remained significant after adjusting for body weight and serum LDL levels (Table 1). Thus, the protective effects of hepcidin deficiency on atherosclerosis appear to be independent of serum iron levels, body weight and lipid levels. These findings implicate reduced macrophage iron as the primary determinant of the protective effects of hepcidin deficiency against atherosclerosis.

Table 1. The protective effect of hepcidin deficiency on the development of atherosclerosis is independent of serum iron levels.

ProSense Near-Infrared Fluorescence Signal (N=50)	Bivariable Analysis		Multivariable Analysis*
Variable	Beta Coefficient† (95% CI)	p-value	Beta Coefficient† (95% CI)	p-value
Hepcidin Deficiency	−0.66 (−0.88 to −0.43)	<0.001	−0.66 (−0.92 to −0.41)	<0.001
Serum Iron (for every 50 μg/dL increase)	0.051 (0.010-0.093)	0.017	0.054 (0.009-0.099)	0.019

^*Multivariable analysis was adjusted for weight and serum LDL levels.

^†Beta coefficient describes the percent change in ProSense fluorescence signal attributable to either hepcidin deficiency or to an increase in serum iron of 50 μg/dL. Negative beta coefficients represent a protective effect against atherosclerosis.

CI, confidence interval

Increased Serum Iron Does Not Prevent Atherosclerosis

Additional studies were performed to confirm that the effects of hepcidin deficiency on atherosclerosis were not mediated by increased serum iron. Hamp^+/+/ Ldlr^−/− mice were fed a high-fat diet and were treated with either intraperitoneal injections of iron dextran or dextran alone. At the end of 21 weeks, the average serum iron in iron-injected LDLR-deficient mice was higher than that in dextran-treated mice (268±10 μg/dL vs 131±14 μg/dL, p<0.0001; Figure 2A). The levels of both hepatic iron and splenic macrophage iron were markedly increased in iron-injected LDLR-deficient mice compared to control LDLR-deficient mice, as determined using Prussian blue stain (not shown). As with Hamp^−/−/ Ldlr^−/− mice, the body weight of iron-treated Ldlr^−/− mice was decreased compared to control Ldlr^−/− mice (Figure 2B). In addition, iron-treated Ldlr^−/− mice had lower fasting serum LDL levels than control animals (Figure 2C). Compared to LDLR-deficient mice treated with dextran, Ldlr^−/− mice treated with iron exhibited no significant difference in atherosclerosis, as evidenced by aortic ProSense signal and Oil Red O staining (Figure 2D-G). The observation that iron loading did not inhibit atherosclerosis further supports the hypothesis that the protective effects of hepcidin deficiency against atherosclerosis are mediated by the resulting decrease in macrophage iron, rather than increased serum iron.

Figure 2.

Iron-loading in Ldlr^−/− mice did not decrease atherosclerosis. (A) Compared to mice treated with dextran alone, intraperitoneal injection of iron dextran into Ldlr^−/− mice produced a greater than two-fold increase in the level of serum iron (p<0.0001, n=6 and 7 respectively). (B) There was a 19% reduction in final body weight (p=0.0003) and (C) a 37% decrease in fasting LDL levels in iron-treated Ldlr^−/− mice compared to controls (p=0.013). (D) There was no difference in macrophage activity in the aortas of iron-treated, compared to control-treated, Ldlr^−/− mice as determined using ProSense near-infrared fluorescence (three representative pairs of aortas are shown). (E) Quantification of ProSense signal revealed no difference in macrophage activity (p=0.12). (F) There was no difference in lipid accumulation in the aortas of iron-treated Ldlr^−/− mice compared to control mice as determined using Oil Red O stain. Aortas from the same three pairs of mice as in (D) are shown. (G) There was also no difference in the area and intensity of aortic staining with Oil Red O between iron-injected and control mice (p=0.16).

Hepcidin Deficiency is Associated with a Reduced Pro-Inflammatory Phenotype in Macrophages

To determine whether hepcidin deficiency alters macrophage phenotype, aortas were harvested from Hamp^+/+/ Ldlr^−/− and Hamp^−/−/ Ldlr^−/− mice and enzymatically digested for flow cytometric analysis. As previously described, leukocytes were identified as CD45⁺ and macrophages as CD45⁺CD11b⁺ (Supplemental Figure IV).^27,28 After treatment with Calcein-AM, CD45⁺CD11b⁺ aortic macrophages from Hamp^−/−/ Ldlr^−/− mice had higher fluorescence than those of Hamp^+/+/ Ldlr^−/− mice, indicating lower intracellular levels of labile iron (Figure 3A). The lower level of intracellular iron was associated with a 33% relative reduction in the percentage of CD38 positive cells (identified as an important M1 pro-inflammatory macrophage marker)³³ in the aortic macrophages of Hamp^−/−/ Ldlr^−/− mice compared to Hamp^+/+/ Ldlr^−/− mice (8.2±0.3 vs 12.3±0.9%, p=0.001; Figure 3B). No difference in the percentage of CD206 positive cells (an M2 alternative macrophage marker) between the aortic macrophages of Hamp^−/−/ Ldlr^−/− mice and Hamp^+/+/ Ldlr^−/− mice was observed (Figure 3C). To confirm that hepcidin deficiency results in a decreased M1 phenotype in macrophages, Hamp^+/+/ Ldlr^−/− and Hamp^−/−/ Ldlr^−/− mice were treated with intraperitoneal thioglycollate. Peritoneal macrophages were isolated, and the levels of iNOS (M1 marker) and arginase-1 (Arg1, M2 marker) were determined by immunoblot. Peritoneal macrophages from Hamp^−/−/ Ldlr^−/− mice had a reduced ratio of iNOS to Arg1 protein compared to that in the peritoneal macrophages from Hamp^+/+/ Ldlr^−/− mice (Figure 3D-E). Expression of CD11b (a pan-macrophage marker) was similar in the macrophages from the two groups of mice. These findings support the hypothesis that hepcidin deficiency with reduced macrophage iron content is protective against atherosclerosis due to a decrease in the pro-inflammatory M1 phenotype of macrophages.

To model hepcidin deficiency in vitro, THP1 human macrophages were treated with the iron chelator deferiprone (DFP). Cells were then grown in the presence or absence of LPS to induce a pro-inflammatory phenotype. LPS treatment was associated with a greater than 18-fold increase in iNOS mRNA levels. In contrast, iNOS mRNA levels were not increased in cells treated with LPS and DFP (Figure 3F), showing that treatment with an iron chelator inhibited the LPS-induced M1 phenotype. Furthermore, treatment of iron-treated THP1 cells with DFP reduced oxidized LDL (oxLDL) uptake by 74% (Figure 3G-H, p=0.001). OxLDL is known to induce expression of the ATP-binding cassette (ABC) transporter ABCA1 (Figure 3I), which removes lipid from macrophages and allows for reverse cholesterol transport to the liver.^34,35 In the presence of oxLDL, treatment of THP1 cells with iron reduced ABCA1 mRNA levels. Compared to iron-treated cells, DFP-treated THP1 cells had 60% higher ABCA1 mRNA levels (p<0.0001). These findings suggest that macrophage iron depletion, similar to that seen in mice with hepcidin deficiency, reduces the pro-inflammatory M1 phenotype of macrophages. In addition, macrophage iron depletion is associated with increased ABCA1 mRNA expression and reduced overall macrophage uptake of oxLDL; the latter is essential for foam cell formation in atherosclerosis.

Discussion

In this study, we report that genetic deletion of Hamp, the gene encoding hepcidin, inhibits the development of atherosclerosis in a murine model. Hamp^−/−/ Ldlr^−/− mice had a greater than two-fold increase in serum iron, but reduced aortic macrophage iron and no detectable iron in splenic macrophages. Hamp^−/−/ Ldlr^−/− mice fed a high-fat diet had a 45% decrease in aortic macrophage activity compared to control Ldlr^−/− mice. Lipid accumulation was also reduced in the aortas of hepcidin-deficient mice. In addition to increased serum iron levels, Hamp^−/−/ Ldlr^−/− mice had decreased body weight and decreased LDL levels. Multivariable linear regression models suggested that the benefits of hepcidin deficiency were not due to differences in serum iron levels, body weight or LDL levels. To further investigate the possible contribution of increased serum iron to decreased atherosclerosis, Hamp^−/−/ Ldlr^−/− mice were injected with iron dextran, although we acknowledge that injection with iron dextran can also increase macrophage iron. As was previously reported by Kautz et al, who performed similar studies using the ApoE-deficient mouse model of atherosclerosis,¹⁴ treatment of Ldlr^−/− mice with iron dextran did not protect against atherosclerosis. Although increasing macrophage iron does not promote atherosclerosis, perhaps because excess macrophage iron may be effectively chaperoned by ferritin,¹⁴ our findings suggest that decreasing the level of intracellular iron in macrophages by targeting hepcidin inhibits atherosclerosis.

Atherosclerosis is a chronic inflammatory disease in which macrophages play an essential role in plaque formation and progression. Macrophages actively take up oxidized LDL thereby generating foam cells; these foam cells contribute to the pro-inflammatory microenvironment of plaques by secreting numerous cytokines, reactive oxygen species, and proteases.³⁶ Macrophage apoptosis accelerates the development of lipid-rich necrotic cores and plaque instability.³⁷ Although macrophages exhibit multiple heterogeneous phenotypes, two major classes that have been identified are pro-inflammatory (M1) macrophages and anti-inflammatory (M2) macrophages.^36,38 M1 macrophages are classically activated by LPS, are enriched in progressing atherosclerotic plaques, and produce high levels of iNOS. M2 macrophages express high levels of arginase, are enriched in regressing plaques, and promote tissue repair.³⁸ We found that the reduced macrophage iron content observed in hepcidin-deficient LDLR^−/− mice was associated with decreased expression of M1 phenotypic markers in both aortic and peritoneal macrophages. Furthermore, treatment of human macrophages in vitro with the iron chelator deferiprone decreased LPS-induced iNOS expression and reduced the uptake of oxLDL relative to iron-treated cells. These results highlight the importance of macrophage iron in promoting a pro-inflammatory, foam cell phenotype.

In addition to the hepcidin-ferroportin axis, intraplaque hemorrhage (IPH) can alter macrophage iron content in atherosclerotic lesions. IPH is commonly observed in advanced atherosclerotic disease in humans. Intimal thickening associated with atherosclerotic lesion formation and progression is thought to create a local environment of hypoxia that induces angiogenesis; however, these new blood vessels are fragile and more permeable than normal and are at increased risk for IPH with associated red blood cell (RBC) extravasation.^39,40 The extravasated RBCs undergo lysis, and free hemoglobin (Hb) subsequently binds to haptoglobin before being taken up by macrophages via the CD163 receptor. Finn and colleagues reported that hemoglobin-stimulated macrophages exhibit increased ABCA1 expression and reduced cholesterol loading as well as increased ferroportin expression and reduced intracellular iron,⁴¹ similar to the macrophages in hepcidin-deficient mice. Macrophages in areas of IPH display a non-foam phenotype that is associated with liver X receptor (LXR)-mediated increased ferroportin expression and iron export as well as enhanced cholesterol efflux with increased ABCA1 expression.^42-44 The potential atheroprotective qualities of these CD163-positive macrophages remain controversial.⁴⁵ Although LDLR-deficient mice develop atherosclerosis, they do not develop IPH,⁴⁰ and are therefore not a suitable model to investigate the effects of free hemoglobin in plaques. However, the results of this study show that hepcidin deficiency is associated with reduced intracellular macrophage iron and a non-foam cell phenotype that is atheroprotective. The results provide direct evidence of a causal link between decreased hepcidin, decreased macrophage iron and decreased atherosclerosis in a murine model.

Hepcidin gene expression is induced by increased levels of serum iron and by inflammation (reviewed in ¹⁶). High levels of iron increase hepcidin by stimulating the bone morphogenetic protein (BMP) signal transduction pathway. In previous studies, inhibition of BMP signaling was shown to decrease atherosclerosis in mouse models of atherosclerosis. Saeed and colleagues treated ApoE^−/− mice with LDN-193189, a small molecule inhibitor of the BMP type 1 receptor, and observed significantly reduced vascular macrophage accumulation and atherosclerotic lesion formation.⁴⁶ Derwall et al observed that Ldlr^−/− mice treated with either LDN-193189 or with ALK3-Fc, a stabilized receptor that binds and sequesters BMP ligands, also had decreased atherosclerosis.⁴⁷ The mechanism by which inhibition of BMP signal transduction decreases atherosclerosis has not been established; BMP signaling exerts many direct effects on the vasculature.⁴⁸ The results of the present study suggest that the beneficial effects of inhibiting the BMP signaling pathway on the development of atherosclerosis may be a result of decreased hepcidin levels. Because BMP signaling has critical roles in maintaining the normal function of many host tissues, chronic inhibition of BMP signal transduction, as a potential approach to the treatment of atherosclerosis, might have a broad range of adverse effects. In contrast, specific targeting of the hepcidin-ferroportin pathway may represent a more selective approach to the treatment of atherosclerosis.

Inflammation induces expression of hepcidin by increasing serum levels of IL-6 and IL-1β.⁴⁹ Previous studies showed that decreased signaling through the IL-6 and IL-1β cytokine pathways ameliorates atherosclerosis in humans. Genetic studies, using Mendelian randomization, showed that the Asp358Ala variant of the IL-6 receptor is associated with decreased risk of coronary artery disease.^50,51 The Asp358Ala IL-6 receptor variant reduces IL-6 signaling in hepatocytes, monocytes and macrophages and results in decreased production of downstream acute-phase proteins, including C-reactive protein and fibrinogen. Decreased IL-6 receptor mediated signaling would be expected to decrease hepcidin levels, suggesting that the beneficial effects of decreased IL-6 signaling on atherosclerosis might be mediated by increased macrophage ferroportin, and decreased macrophage iron. A limitation to the use of inhibitors of IL-6 signaling as a treatment for atherosclerosis is the associated increased risk of infection.

After 21 weeks on the high-fat diet, the body weight of Hamp^−/−/ Ldlr^−/− mice was 8% less than that of Hamp^+/+/ Ldlr^−/− mice. A small, but statistically significant decrease in body weight was noted by Lesbordes-Brion and colleagues in their initial description of mice with a targeted disruption of the hepcidin gene.²⁰ The decrease in body weight may be secondary to the adverse effects of iron deposition in a range of tissues, including heart, lungs, pancreas, kidneys, muscle and joints. In addition, compared to Hamp^+/+/ Ldlr^−/− mice, Hamp^−/−/ Ldlr^−/− mice had a 37% decrease in fasting serum LDL. Patients with hemochromatosis caused by the C282Y mutation in the HFE gene were also reported to have decreased LDL levels.⁵² The mechanism by which hemochromatosis, with its associated hepcidin deficiency and iron overload, results in decreased serum LDL levels is unknown.

In summary, hepcidin deficiency resulted in significantly decreased aortic macrophage activity and atherosclerosis in LDLR-deficient mice and was associated with a reduced inflammatory macrophage phenotype and decreased uptake of oxidized LDL. Targeting hepcidin and its interaction with ferroportin may represent a novel approach to the treatment of cardiovascular disease.

Highlights.

• Genetic deletion of the gene encoding hepcidin reduced aortic macrophage activity and inhibited the development of atherosclerosis in LDLR^−/− mice fed a high fat diet.

• The protective effects of hepcidin deficiency on the development of atherosclerosis were associated with a reduced pro-inflammatory phenotype in aortic macrophages and occurred independently of its effects on serum iron.

• Decreasing the level of intracellular iron in macrophages by targeting hepcidin may be a novel approach in the treatment of atherosclerosis.

Abbreviations

ALK3

activin-like kinase 3

ApoE

apolipoprotein E

BMP

bone morphogenetic protein

CI

confidence interval

DFP

deferiprone

HAMP

hepcidin antimicrobial peptide

HDL

high-density lipoprotein

iNOS

inducible nitric oxide synthase

IPH

intraplaque hemorrhage

LDL

low-density lipoprotein

LDLR

low-density lipoprotein receptor