Pharmacologic Suppression of Hepcidin Increases Macrophage Cholesterol Efflux and Reduces Foam Cell Formation and Atherosclerosis

Abstract

Objectives

We recently reported that lowering of macrophage free intracellular iron increases expression of cholesterol efflux transporters ABCA1 and ABCG1 by reducing generation of reactive oxygen species. In this study, we explore whether reducing macrophage intracellular iron levels via pharmacologic suppression of hepcidin can increase macrophage-specific expression of cholesterol efflux transporters and reduce atherosclerosis.

Methods and Results

To suppress hepcidin, increase expression of the iron exporter ferroportin (FPN), and reduce macrophage intracellular iron, we used a small molecule inhibitor of BMP signaling, LDN 193189 (LDN). LDN (10 mg/kg i.p. bid) was administered to mice and its effects on atherosclerosis, intracellular iron, oxidative stress, lipid efflux, and foam cell formation were measured in plaques and peritoneal macrophages. Long-term LDN administration to Apo E (-/-) mice increased ABCA1 immunoreactivity within intraplaque macrophages by 3.7-fold (n=8; p=0.03), reduced oil-red-o positive lipid area by 50% (n=8; p=0.02) and decreased total plaque area by 43% (n=8; p=0.001). LDN suppressed liver hepcidin transcription and increased macrophage FPN, lowering intracellular iron and hydrogen peroxide production. LDN treatment increased macrophage ABCA1 and ABCG1 expression, significantly raised cholesterol efflux to ApoA-1 and decreased foam cell formation. All preceding LDN-induced effects on cholesterol efflux were reversed by exogenous hepcidin administration, suggesting that modulation of intracellular iron levels within macrophages as the mechanism by which LDN triggers these effects.

Conclusion

These data suggest that pharmacologic manipulation of iron homeostasis may be a promising target to increase macrophage reverse cholesterol transport and limit atherosclerosis.

Atherosclerosis progresses through intracellular lipid accumulation within macrophages leading to foam cell formation and necrotic core growth^{1, 2}. Although current clinical strategies have focused on cholesterol lowering as a way to decrease lipid retention in the arterial wall, increasing macrophage lipid efflux has been suggested to be another promising strategy to limit foam cell formation and atherosclerosis³. Efflux of intracellular lipid occurs primarily through ATP-binding cassette (ABC) transporters, ABCA1 and ABCG1, resulting in removal of lipid from macrophages and reverse cholesterol transport to the liver through plasma high-density lipoproteins^4-6. Genetic deletion of ABCA1 and ABCG1 augments foam cell formation while overexpression of these genes in macrophages slows progression of atherosclerotic lesions in animal models^{4, 7}. While elegant genetic studies lend important insights into disease mechanisms, their potential for clinical translation is limited since selective gene manipulation within humans is not currently a viable option. To utilize the therapeutic potential of increasing macrophage cholesterol efflux for the prevention of atherosclerosis in humans, new pharmacologic means to increase macrophage expression of lipid efflux transporters are needed.

We recently described within areas of intraplaque hemorrhage in post mortem human atherosclerotic plaques a specific subtype of macrophages which we termed M(Hb). These macrophages resist foam cell formation both in vivo and in response to exogenous cholesterol loading, have increased expression of ABCA1 and ABCG1, and anti-oxidative characteristics. We showed that the anti-oxidative properties of M(Hb) are causal in increased expression of ABCA1 and ABCG1 and originate from a reduction in intracellular free iron available for electron donation for reactive oxygen species (ROS) formation⁸. M(Hb) have reduced intracellular free iron due to increased expression of a free iron exporter, ferroportin (FPN). Our data suggest that reducing intracellular free iron levels within macrophages by increasing expression of macrophage FPN may be a promising strategy to increase expression of cholesterol efflux transporters.

FPN is the only known mammalian free iron exporter expressed by macrophages and it is systemically degraded through ubiquitination after binding to a hepatic hormone, hepcidin⁹. The promoter elements of hepcidin are activated by SMAD 1/5/8 transcription factors, which are in turn activated through bone morphogenetic protein (BMP) signaling¹⁰. We used a novel small molecule inhibitor of BMP signaling, LDN 193189, which prevents the activation of SMAD 1/5/8 to suppress hepatic hepcidin production and increase expression of FPN within macrophages¹¹. We explored the effects of this strategy on mouse atherosclerosis, macrophage intracellular iron levels, oxidative stress, lipid efflux, and foam cell formation. Our findings reveal that suppressing hepcidin by inhibiting BMP signaling through LDN 193189 (LDN) significantly increases expression of ABCA1 and ABCG1 and lipid efflux by macrophages, which is associated with reduced foam cell formation and atherosclerosis in the Apo E (-/-) mouse model.

Methods

Chemicals

LDN-193189 (4-[6-(4-piperazin-1-ylphenyl)pyrazolo[1,5-a]pyrimidin-3-yl]quinoline) was synthesized as previously described¹². The vehicle was 2% (wt/vol) (2-hydroxypropyl)-β-cyclodextrin in PBS, pH 7.4. Control animals received vehicle alone.

Animals and experimental protocols

The Institutional Animal Care and Use Committee at Emory University approved all animal protocols. All animal experiments were conducted according to the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Male, 12-14 weeks old C57BL/6J mice, and male, 10-12 weeks old Apolipoprotein E- knockout (Apo E -/-) mice on a C57BL/6J background, were purchased from Jackson Laboratories (Bar Harbor, ME).

Atherosclerosis

To investigate the effect of long-term hepcidin suppression on atherosclerosis, we placed Apo E -/- mice into two groups (n=8, per group): control animals receiving vehicle i.p. bid × 10 weeks and LDN treated animals receiving LDN 10 mg/kg i.p. bid × 10 weeks. After starting LDN or control, all mice were placed on an ad libitum chow diet (Harlan Teklad, Madison, WI, USA) for four days and on day 5 animals were switched to an atherogenic diet (1.25% cholesterol, 21% fat, Harlan Tekland, Madison, WI, USA) which was continued until sacrifice. At the end of the 10 week treatment period, the left ventricular ejection fraction was measured by the VEVO 2100 high resolution echocardiogram system (Visual Sonics, Toronto, Canada) and blood was collected by cardiac puncture for hematocrit, serum iron and lipid measurements. Then mice were perfused with normal saline and fixed with 4% paraformaldehyde (PFA) through the left ventricle of the heart. After fixation, the heart, aortic arch and descending aorta were collected for analysis of atherosclerotic plaque burden and immunohistochemistry.

LDN efficacy studies

To establish the efficacy of LDN to suppress liver hepcidin we divided C57BL/6J mice into two groups: control animals receiving vehicle through intraperitoneal (i.p.) injection every 12 hours (bid) × 4 days and LDN treated animals receiving a previously efficacious dose¹² of 10mg/kg i.p. bid × 4 days. On day five, blood was collected for serum iron measurements and liver samples were taken to measure hepcidin expression.

Peritoneal macrophage isolation

Control animals received vehicle i.p. bid × 4 days and LDN treated animals received LDN 10mg/kg i.p. bid × 4 days. On day 5, all mice underwent peritoneal lavage. Peritoneal cells were plated for overnight incubation with one set of cells from LDN treated animals exposed to 700nM mouse hepcidin (Peptides International, Louisville, KY.) to evaluate for reversal of LDN induced effects⁹. The following day all non adherent cells were removed and, after washing 3× with PBS, we were able to obtain >85% macrophages as identified by a fluorescent anti-mouse f4/80 Alexaflour 488 antibody (eBioscience) (supplemental figure I), consistent with previous reports¹³. Peritoneal macrophages were scraped off plates for molecular and cellular analysis.

Foam cell formation

C57BL/6J mice were placed in the following three groups: 1) control animals receiving vehicle i.p. bid × 4 days; 2)LDN treated animals receiving LDN 10mg/kg i.p. bid × 4 days; and 3) LDN plus hepcidin treated animals receiving LDN 10 mg/kg i.p. bid × 4 days and 25 ug (dissolved in 100 ul PBS) of mouse hepcidin i.p. once daily on days 3 and 4. The 25 ug dose of mouse hepcidin has been previously validated to induce hypoferremia by FPN degradation¹⁴. To induce peritoneal foam cell formation, all mice were given an i.p. injection of 500 micrograms of 1mg/ml oxidized LDL (Intracel, Fredrick, MD) on day 3. On day 5, all mice underwent peritoneal lavage. The peritoneal cells were plated overnight in serum free media with 1% Nutridoma (Boehringer). On the following day, all non adherent cells were removed. After washing 3× with PBS, cells were fixed in 4% paraformaldehyde and stained with Oil Red O and Hemotoxylin and Eosin. Cytoplasmic Oil Red O staining was quantified by color density analysis using iVision software (Biovision).

Cholesterol Loading and Efflux Assays

C57BL/6J mice were placed into the following three groups: 1) control animals receiving vehicle i.p. bid × 4 days; 2) LDN treated animals receiving LDN 10mg/kg i.p. bid × 4 days; and 3) LDN plus hepcidin treated animals receiving LDN 10 mg/kg i.p. bid × 4 days and 25 ug (dissolved in 100 ul PBS) hepcidin i.p. once a day on days 3 and 4. On day 5, peritoneal cells were removed and loaded with 30 ug/ml of o× LDL for 48 hours. For cholesterol uptake studies, cells were then analyzed for cholesterol content by enzymatic assay as described below. For efflux studies, after this incubation period, cells were washed twice in PBS and apoAI-mediated cholesterol efflux studies were immediately performed by adding fresh RPMI medium without Nutridoma with or without 20 μg/ml of apoAI (BioVision) for 24 h. At the end of this incubation, intracellular lipids were extracted in hexane/isopropanol, dried under nitrogen and free cholesterol, and total cholesterol and phospholipids were measured by enzymatic assays (Calbiochem). Esterified cholesterol was measured as the difference between total and free cholesterol. Cellular proteins were collected by digestion in NaOH and measured by Bradford assay (BioRad). The percent change of intracellular cholesterol amounts in the presence of apoAI relative to apoAI-free medium was expressed according to the following equation: percent decrease in cellular cholesterol = {[(cellular cholesterol) RPMI − (cellular cholesterol)ApoAI] ÷ [cellular cholesterol]RPMI} × 100.

RNA isolation and quantitative PCR

Total cellular RNA was isolated from macrophages using TRIZOL (Invitrogen, Carlsbad, CA) and was reverse transcribed to create cDNA. cDNA was quantified by quantitative polymerase chain reaction on a StepOne Plus (Applied Biosystems). The primer sequences for all studied genes are listed in supplemental table I. Messenger RNA levels are normalized to GAPDH.

Western Blot

Western blot was conducted for ABCA1 (Abcam antibodies, Cambridge, MA) and ferroportin (Alpha Diagnostics, San Antonio, TX) as previously described⁸. Protein samples are normalized to alpha Tubulin (Cell Signaling Technology, Danvers, MA). Densitometry was performed as previously described¹⁵.

Measurement of macrophage intracellular free iron by calcein florescence

Intracellular macrophage iron was measured by using calcein florescence, an assay that quenches calcein florescence by free iron, as described previously¹⁶. In brief, macrophages were stained with 0.5 uM calcein (Calbiochem) and then analyzed by a BD SR II flow cytometer and FlowJo software 7.5. Cellular florescence is inversely related the free iron levels.

Measurement of macrophage hydrogen peroxide production using Amplex Red

Macrophages were incubated in a 96 well plate with 50ul of 100uM Amplex red/HRP solution (Invitrogen). After incubation at room temperature for 24 hours, protected from light, fluorescence was measured at excitation 530nm and emission 590nm. Cell viability using propidium iodide staining and flow cytometry was conducted at the end of 24 hours on cells from each group and demonstrated no differences between groups (control 87.5±1.5% viable cells (i.e. PI exclusion) in control cells versus 87.3±1% in LDN cells, p=ns, n=4 experiments per group).

Measurements of hematocrit, serum iron and lipids

Approximately 1.5 ml of mouse blood was collected in heparin syringes by cardiac puncture. 0.5 ml of blood was placed in separate tube for hematocrit measurement, and the remaining sample was centrifuged at 5,000 RPM for 15 min to obtain serum. Hematocrit, serum iron, total lipid and triglycerides were measured commercially (Antech Diagnostics, Atlanta, GA).

Atherosclerotic lesion analysis and Immunohistochemistry

The heart and aortic arch were removed en bloc and frozen for tissue analysis. To obtain frozen sections, tissue was embedded in OCT, frozen in liquid nitrogen and stored at -80 C. For morphometric lesion analysis, cryosections beginning at the base of the aortic root were obtained. Consecutive cross sections at the level of the sino-tubular junction were stained with hematoxylin and eosin, Movat pentachrome, and oil red o, as previously described¹⁷. Immunohistochemistry was done using primary rat anti-mouse Mac 3 (1:100, BC Pharmagen) and primary mouse monoclonal ABCA1 antibody (1:200, Lifespan Technologies Inc.) using the avidin-biotin-peroxidase complex (ABC) method. Specificity of the antibody was confirmed by demonstrating lack of immunostaining of macrophages from aortic lesions from LDLR (-/-) transplanted with bone marrow from ABCA1/ABCG1 (-/-) mice⁴ (courtesy of Alan Tall (Columbia University, NY) (supplementary figure II). Cross sections were imaged using an Olympus microscope and stained area was quantified by segmentation color-threshold analysis using morphometry software (IP Lab, Scanalytics, Rockville, MD). The lesion area in the descending aorta was analyzed as previously described¹⁸. For analysis of liver iron, Perl's iron stain was conducted on livers from 10 week treated control and LDN treated Apo E (-/-) mice.

Statistical Analysis

Data are expressed as a mean with +/- standard error bars. For comparisons between two groups for continuous variables, a one-way ANOVA test was used by JMP software. p < 0.05 is considered statistically significant.

Results

Effect of LDN on ABCA1 immunoreactivity in plaque macrophages, oil red o positive lipid area, total atherosclerotic lesion area and plaque progression

To evaluate the impact of LDN on atherosclerosis, Apo E (-/-) mice on a high cholesterol diet were treated with LDN for ten weeks. Atherosclerotic plaques in the sino-tubular junction from LDN and vehicle (PBS) treated Apo E (-/-) mice were localized for histological analysis. Oil red o positive lipid cell area was significantly reduced in LDN treated animals by 50% (0.035mm² vs. 0.07mm²; n=8; p=0.024) and total plaque area at the sinotubular junction was reduced by 43% (0.35mm² vs. 0.20mm²; n=8; p=0.001) in the LDN group in comparison to vehicle (figure 1A-C). The lesion area in the descending aorta also decreased by 45% with LDN treatment (4.6% vs. 2.5%; n=8; p=0.002; fig 1D, E). ABCA1 immunoreactivity was measured within intraplaque Mac-3 positive macrophages regions and showed a 3.7 fold increase (n=8; p=0.03) with LDN treatment in comparison to vehicle (figure 1F, G). Moreover, atherosclerotic plaques were classified by severity, as previously described¹⁹, and mice that received LDN had decreased number of advanced lesions with necrotic cores in comparison to vehicle (figure 1H).

Figure 1. Effect of LDN on atherosclerosis development in Apo E (-/-) mice.

A, Representative 40× and 400× photomicrographs of plaques at the sinotubular junction from Apo E -/- mice treated with PBS (control) or LDN × 10 weeks, stained with hematoxylin and eosin (H&E) or Movat Pentachrome or Oil Red O. B, Quantitation of atherosclerotic plaque area at the sinotubular junction as visualized by Movat Pentachrome stain from mice treated with PBS (control) or saline (n=8, per group). C, Quantitation of intraplaque lipid area at the sinotubular junction as visualized by oil red o staining from mice treated with PBS (control) or LDN (n=8, per group). D-E, representative images and quantitation of the percentage of plaque area in the descending aorta of mice treated with PBS (control) or LDN (n=8, per group). F, representative 100× photomicrographs of immunohistochemical staining for the macrophage marker Mac-3 and the lipid efflux transporter ABCA1. G, percentage of ABCA1 immunoreactivity within a Mac 3 positive area in plaques at the sinotubular junction from Apo E -/- mice treated with PBS (control) or LDN (n=8, per group). H, plaques with characteristics of increasing histological severity were identified and quantitated at the sinotubular junction of mice treated with PBS (control, black bars) or LDN (orange bars, n=8, per group).

Effect of long-term LDN administration on body weight, lipids, hematocrit, cardiac function and serum and liver iron

10 week LDN treatment resulted in no significant differences in comparison to controls in weight gain, total cholesterol, triglycerides, hematocrit and cardiac ejection fraction (see table 1). Serum iron levels were significantly higher with LDN treatment in comparison to control (130 ug/dl vs. 178 ug/dl, p<0.05) as expected given the increased expression of macrophage FPN. Despite these higher levels of iron, no differences in liver iron as detected by Perl stain were seen in control versus LDN treated animals after ten weeks of treatment (data not shown).

Table 1.

Weight gain, serum total cholesterol levels, serum triglyceride levels, hematocrit, serum iron and cardiac ejection measurements from Apo E -/- mice treated with PBS (control) or LDN × 10 weeks.

Endpoint	Control (n=8)	LDN (n=8)	p value
Weight gain (g)	2.89 (+/- 1.48)	1.57 (+/- 1.0)	n.s
Total Cholesterol (mg/dL)	1110 (+/- 88)	1085 (+/-152)	n.s
Triglycerides (mg/dL)	128 (+/- 22)	109 (+/- 29)	n.s
Hematocrit (g/dL)	38.4 (+/- 2.4)	38.7 (+/- 2.6)	n.s
Serum Iron (ug/dL)	130 (+/- 6)	178 (+/- 22)	p = 0.02
Ejection Fraction (%)	68.1 (+/- 1.6)	60.6 (+/- 5.6)	n.s

Effect of inhibiting BMP signaling by LDN on liver hepcidin production, FPN expression, intracellular iron, and hydrogen peroxide production in peritoneal macrophages

In order to investigate whether manipulation of macrophage intracellular iron is an important mechanism by which LDN decreases atherosclerosis, we first confirmed the efficacy of LDN to suppress liver hepcidin as reported previously^{10, 20}. 10 mg/kg of LDN administered i.p. bid for 4 days significantly suppressed liver hepcidin mRNA by greater than ten-fold in comparison to vehicle treatment (figure 2A). To determine if hepcidin suppression increased FPN, we isolated peritoneal macrophages from LDN treated mice and compared FPN expression to controls by western blot. FPN was significantly increased after LDN treatment (figure 2B, C). This upregulation of FPN by LDN was ablated by exogenous hepcidin (figure 2B, C), consistent with the known effect of hepcidin on FPN expression. Next, we evaluated the impact of LDN treatment on intracellular iron and hydrogen peroxide production. Peritoneal macrophages isolated after LDN treatment showed reduced intracellular iron and hydrogen peroxide production (figure 2E, F). The effects of LDN on intracellular iron and hydrogen peroxide production were reversed with exogenous hepcidin (figure 2E, F).

Figure 2. LDN inhibits systemic iron metabolism and affects macrophage intracellular iron, reactive oxygen species, and ABC transporter expression.

A, Expression of liver hepcidin in mice after treatment with PBS (control) or LDN (n=7 per group). B, Western blots showing FPN in mouse peritoneal macrophages after treatment with PBS (control) or LDN or LDN + hepcidin. Alpha tubulin is shown as an internal control. C, Relative densitometry of FPN/alpha tubulin western blots in mouse peritoneal macrophages after treatment with PBS (control) or LDN or LDN + hepcidin (n=3, per group). D, Serum iron measurements in mice treated with PBS (control) or LDN (n=7, per group). E, Calcein fluorescence histograms of peritoneal macrophages from mice treated with PBS (control, black line) or LDN (orange line) or LDN + hepcidin (blue line). Calcein florescence is inversely proportional to intracellular iron. This is a representative image from four separate experiments. F, Hydrogen peroxide production, as quantitated by Amplex Red in peritoneal macrophages from mice treated with PBS (control) or LDN or LDN + hepcidin (n=4, per group). G, ABCA1 and ABCG1 expression in peritoneal macrophages from mice treated with PBS (control) or LDN (n=4, per group). H, Western blots showing ABCA1 in mouse peritoneal macrophages after treatment with PBS (control) or LDN or LDN + hepcidin. Alpha tubulin is shown as an internal control. I, Relative densitometry of FPN/alpha tubulin western blots in mouse peritoneal macrophages after treatment with PBS (control) or LDN or LDN + hepcidin (n=4, per group)

Effect of LDN on ABCA1 and ABCG1 expression, cholesterol efflux, and foam cell formation

To determine if the anti-oxidative effects produced by lowering intracellular iron in macrophages by LDN impact lipid efflux we measured the expression of ABCA1 and ABCG1. Peritoneal macrophages isolated after 4 days of LDN treatment showed a significant 2-fold and 7.3-fold increase in mRNA of ABCA1 and ABCG1, respectively, in comparison to vehicle (figure 2G). Correspondingly, ABCA1 protein levels were increased with LDN treatment and returned to near baseline with exogenous overnight hepcidin treatment (figure 2H, I), indicating the importance of FPN in the mechanism of ABC transporter upregulation. Overnight in-vitro incubation of LDN (at a dose previously shown to inhibit BMP signaling²¹) with isolated peritoneal macrophages had no effect on the expression of ABCA1 and ABCG1 (supplemental figure IIIA), further supporting an iron-related mechanism of ABC transporter expression by LDN. To evaluate if increased expression of ABCA1 and ABCG1 influence lipid efflux, functional experiments were performed to determine the influence of LDN treatment on macrophage cholesterol efflux to ApoAI. We incubated peritoneal macrophages from control and LDN treated mice with oxLDL (30ug/ml) for 48 hours to induce cholesterol accumulation. We subsequently exposed them to ApoAI (20ug/ml) for 24 hours to induce cholesterol efflux and then determined cholesterol levels by enzymatic assay. ApoAI treatment significantly increased cholesterol efflux in macrophages from LDN treated animals as compared to control as indicated by significantly decreased levels of total, free, and esterified cholesterol levels (figure 3A). This increase in lipid efflux by LDN was ablated by concurrent hepcidin treatment (figure 3A). Lastly, to observe the influence of LDN on morphologic foam cell formation we stained lipid loaded macrophages isolated after LDN versus control treatment with oil red o. LDN treated macrophages had a significant reduction in foam cell formation as evident by reduced cytoplasmic oil red o staining in comparison to control (figure IIIB). These non-foam cell foaming effects of LDN were reversed with concurrent exogenous hepcidin (figure 3 C-E). These data suggest that the effects of LDN on macrophage ABC transporter expression and cholesterol efflux are mediated via its effects on macrophage intracellular iron.

Figure 3. Effect of LDN of macrophage cholesterol efflux and foam cell formation.

A, peritoneal macrophages from mice treated with PBS (control) or LDN or LDN + Hepcidin were loaded with ox LDL(30 ug/ ml) for 48 hours and then were incubated with RPMI 1640 medium with or without apoA-1 (20 μg/ml) for 24 hours. Intracellular lipids were determined. Results are the mean of 4 experiments and are expressed as the percent change of intracellular cholesterol amounts in the presence of apoAI relative to apoAI-free medium, calculated as described. *p<0.05 vs. control and LDN+hepcidin. B, quantitation of the percentage of cytoplasm stained with oil red o by color density threshold analysis in ox LDL loaded peritoneal cells of mice treated with PBS or LDN or LDN + Hepcidin (n=4, per group). C, 20× oil red o stained images of peritoneal macrophages isolated after in-vivo oxidized LDL loading from mice treated with PBS (C) or LDN (D) or LDN + Hepcidin (E).

To further investigate whether the lack of foam cell formation observed with LDN treatment was due increased cholesterol efflux rather than to limited lipid uptake, scavenger receptor (SR) expression and lipid uptake were examined in the peritoneal macrophages of LDN treated mice. Quantitative PCR data for both class A (I and II) and class B SR showed no changes in expression of these receptors (supplemental figure 3B) compared to control. In addition, macrophages from control and LDN treated animals were loaded in vitro with oxLDL (30ug/ml) for 48 hours and cholesterol content determined by enzymatic assay. Both total and free cholesterol content increased to a similar degree in macrophages from control and LDN animals exposed to oxLDL as compared to unexposed cells (supplemental figure IIIC). However, this was not the case for esterified cholesterol which increased over 500% in control cells receiving oxLDL but only 44% in cells from LDN treated animals (supplemental figure 3C). These data indicate that while some uptake mechanisms still seem to be active in macrophages from LDN treated animals, those involving production of esterified cholesterol, the hallmark of foam cells, are not as active in macrophages from LDN treated animals as compared to control cells.

Discussion

Atherosclerosis is primarily an inflammatory disease driven by uptake of oxidized LDL into macrophages transforming them into foam cells¹. Current clinical strategies have focused on lipid lowering using primarily the statin class of HMG-CoA reductase inhibitors as a means to prevent atherosclerosis progression. Yet, even in this patient population, measures of cholesterol efflux potential such as HDL remain independent predictors of cardiovascular risk²². Despite abundant experimental data using pharmacologic or genetic means to demonstrate that increasing macrophage cholesterol efflux proteins ABCA1 and ABCG1 might be another promising strategy to prevent atherosclerosis progression^{23, 24}, a limited number of agents to increase macrophage cholesterol efflux are clinical development. Our previous work suggests that modulating intracellular iron levels within macrophages by increasing macrophage FPN may be a promising strategy to increase expression of cholesterol efflux transporters⁸. Here we demonstrate for the first time how pharmacologic alternation of systemic iron metabolism using a compound that suppresses liver hepcidin can alter macrophage cholesterol efflux transporters and reduce atherosclerotic lesion progression.

In this study we show that suppression of liver hepcidin by a BMP signaling inhibitor, LDN, leads to increased expression of FPN, reduced intracellular iron and oxidative stress within peritoneal macrophages. These molecular alterations are associated with increased expression of ABC transporters, ABCA1 and ABCG1. Macrophages from LDN treated mice demonstrated increased lipid efflux and reduced foam cell formation. Further mechanistic data to support a causal role for manipulation of intracellular iron by LDN in ABC transporter expression and foam cell formation were made using hepcidin, a liver peptide know to degrade FPN. Exogenous hepcidin was able to reverse expression of macrophage ABCA1, cholesterol efflux, and the effects of LDN on foam cell formation, suggesting the importance of FPN in these LDN mediated effects. This conceptual framework is depicted in figure 4. In addition, in-vitro incubation of LDN with isolated peritoneal macrophages had no effect on the expression of ABCA1 and ABCG1. Collectively these data suggest that hepcidin must be suppressed in-vivo to increase the expression of ABCA1 and ABCG1 in macrophages.

Figure 4. Summarizing schematic.

A, (1) BMP binds (2) BMP-Receptor in the liver and activates specific transcription factors (3) leading to transcription of hepcidin. (4) Hepcidin degrades FPN in macrophages to cause accumulation of intracellular iron and iron induced ROS, which (5) inhibits expression of ABCA1 and ABCG1 and lipid efflux. B, LDN inhibits BMP signaling to prevent activation of hepcidin transcription factors leading to reduced hepcidin and maintenance of FPN in macrophages, thereby, reducing intracellular iron, ROS and increasing the expression of ABCA1 and ABCG1 to promote lipid efflux and reduce foam cell formation.

We also quantified the expression of lipid scavenger receptors and cholesterol uptake by macrophages to assess if the lack of foam cell formation observed with LDN treatment was due to limited lipid uptake. We found no changes in the expression of lipid uptake receptors and in the quantity of total and free cholesterol after lipid loading. However, esterified cholesterol levels were significantly less in macrophages from LDN treated animals as compared to control animals after cholesterol loading. Although not specifically investigated here, we have previously shown that the effects of lowering intracellular iron triggers ABC transporter activation through induction of Liver X receptor alpha (LXRα) which is known to reduce cholesterol ester formation in macrophages^{8, 25}. Thus, we cannot preclude that the effects of LDN on reducing foam cell formation and atherosclerosis could also be related to this mechanism in addition to increasing reverse cholesterol transport.

To evaluate the effect on atherosclerosis, LDN was administered to Apo E -/-mice on a high cholesterol diet for ten weeks. Results show a significant reduction in intraplaque oil red o positive lipid area, total plaque area and plaque severity, along with elevated ABCA1 immunoreactivity within plaque macrophage rich regions. These findings suggest LDN increases the expression of ABCA1 in macrophages within atherosclerotic plaques leading to limited lesion progression and plaque burden. Further support for our proposed mechanism was made by showing that long-term LDN treatment resulted in no changes in serum total cholesterol and triglycerides in comparison to controls.

We did observe that LDN treated Apo E -/- mice had less weight gain; however, these findings were not statistically significant and may be attributable to increased variability in the cohort. Previous investigators have reported this compound is well tolerated¹². Lastly, although mice treated with LDN had higher serum iron levels related to macrophage iron loss, we did not observe this manifest as iron overload toxicity on hematocrit, liver iron deposition, and cardiac function.

Prior findings from human coronary plaques and in-vitro studies⁸ revealed that reduction in intracellular iron and iron induced oxidative stress by increased FPN, are molecular triggers that promote expression of ABCA1 and ABCG1 within macrophages localized to regions of intraplaque hemorrhage. In this investigation, we have translated these findings by pharmacologically increasing FPN in macrophages through suppressing hepcidin to increase their lipid efflux capacity and reduce foam cell formation.

A limitation inherent to our findings is that LDN does not directly inhibit hepcidin but yields its effect by inhibiting upstream BMP signaling. Prior investigations show an atheroprotective role of inhibiting BMP by transgenic overexpression of matrix Gla protein (MGP), an endogenous inhibitor of BMP signaling, leading to reduced vascular calcification, expression of ICAM / VCAM and inflammation²⁶. Thus, our findings pertaining to limited foam cell formation by augmentation of lipid efflux provide an additional--perhaps more direct--mechanism in the atheroprotective role of inhibiting systemic BMP signaling.

Despite the promising results shown here, it is important to note that before human therapy could be considered using this approach, significantly longer and more comprehensive toxicity studies would need to be performed to specifically define potential effects of short and long term treatment given its potential effects on iron related toxicities and tissue repair^{10, 27, 28}. Given the pivotal role of hepcidin in regulating iron homeostasis, its chronic inhibition could potentially result in an iron overload-like state. Although multiple mechanisms exist within humans to counteract iron-related toxicity including the iron binding proteins ferritin and transferrin, it remains possible that such toxicity could be a limitation to the actual clinical adoption of such as strategy.

In conclusion, we show for the first time that the selective BMP inhibitor LDN can limit foam cell formation and atherosclerosis by reducing macrophage intracellular iron leading to enhanced ABC transporter expression and lipid efflux capacity. The molecular pathways by which iron induced oxidative stress limits the expression of ABCA1 and ABCG1 are beyond the scope of this investigation but clearly merit further study. Nonetheless, the translational findings reported in this investigation emphasize that reducing intracellular iron in macrophages can be applied as potential therapeutic strategy to augment reverse cholesterol transport and limit atherosclerosis.