Author + information
- Received August 16, 2011
- Revision received October 5, 2011
- Accepted October 11, 2011
- Published online January 10, 2012.
- Aloke V. Finn, MD⁎,⁎ (, )
- Masataka Nakano, MD†,
- Rohini Polavarapu, BA⁎,
- Vinit Karmali, MA⁎,
- Omar Saeed, MD⁎,
- XiaoQing Zhao, PhD†,
- Saami Yazdani, PhD†,
- Fumiyuki Otsuka, MD†,
- Talina Davis⁎,
- Anwer Habib, MD⁎,
- Jagat Narula, MD, PhD‡,
- Frank D. Kolodgie, PhD† and
- Renu Virmani, MD†
- ↵⁎Reprint requests and correspondence:
Dr. Aloke V. Finn, Emory University Hospital Midtown, 550 Peachtree Street NE, Atlanta, Georgia 30307
Objectives The purpose of this study was to examine selective macrophage differentiation occurring in areas of intraplaque hemorrhage in human atherosclerosis.
Background Macrophage subsets are recognized in atherosclerosis, but the stimulus for and importance of differentiation programs remain unknown.
Methods We used freshly isolated human monocytes, a rabbit model, and human atherosclerotic plaques to analyze macrophage differentiation in response to hemorrhage.
Results Macrophages characterized by high expression of both mannose and CD163 receptors preferentially exist in atherosclerotic lesions at sites of intraplaque hemorrhage. These hemoglobin (Hb)-stimulated macrophages, M(Hb), are devoid of neutral lipids typical of foam cells. In vivo modeling of hemorrhage in the rabbit model demonstrated that sponges exposed to red cells showed an increase in mannose receptor–positive macrophages only when these cells contained Hb. Cultured human monocytes exposed to Hb:haptoglobin complexes, but not interleukin-4, expressed the M(Hb) phenotype and were characterized by their resistance to cholesterol loading and up-regulation of ATP-binding cassette (ABC) transporters. M(Hb) demonstrated increased ferroportin expression, reduced intracellular iron, and reactive oxygen species (ROS). Degradation of ferroportin using hepcidin increased ROS and inhibited ABCA1 expression and cholesterol efflux to apolipoprotein A-I, suggesting reduced ROS triggers these effects. Knockdown of liver X receptor alpha (LXRα) inhibited ABC transporter expression in M(Hb) and macrophages differentiated in the antioxidant superoxide dismutase. Last, LXRα luciferase reporter activity was increased in M(Hb) and significantly reduced by overnight treatment with hepcidin. Collectively, these data suggest that reduced ROS triggers LXRα activation and macrophage reverse cholesterol transport.
Conclusions Hb is a stimulus for macrophage differentiation in human atherosclerotic plaques. A decrease in macrophage intracellular iron plays an important role in this nonfoam cell phenotype by reducing ROS, which drives transcription of ABC transporters through activation of LXRα. Reduction of macrophage intracellular iron may be a promising avenue to increase macrophage reverse cholesterol transport.
Macrophages are the major inflammatory cells involved in the progression of atherosclerosis. Microenvironment directs these cells into morphologically and functionally distinct phenotypes, which suggests that cellular events within the complex milieu of atherosclerosis influence macrophage differentiation. Some macrophages ingest lipoprotein particles and become foam cells, whereas the character and function of others are less clear. At least 2 distinct subtypes of macrophages (so-called M1 and M2) have been shown to exist within atherosclerotic plaques (1). Although the stimulus for alternative activation of macrophages within human atherosclerosis is thought to be mediated through the action of cytokine interleukin (IL)-4 (2), the precise signals that trigger macrophage differentiation remain incompletely understood.
Intraplaque hemorrhage is regarded as a potentially important inflammatory stimulus capable of promoting the influx of macrophages into atherosclerotic lesions (3). The pro-oxidant environment within the plaque helps to promote erythrocyte lysis where free hemoglobin (Hb) is bound by the plasma protein haptoglobin (Hp), forming Hb:Hp (4). The CD163 receptor, exclusively expressed in monocyte/macrophages, binds to Hb:Hp, mediating its endocytosis, and the heme subunit of Hb is degraded by the heme oxygenase enzymes (5). Previous data suggest a critical role for binding of Hb:Hp to CD163 in directing the anti-inflammatory responses of macrophages including the up-regulation of IL-10 (6). Boyle et al. (7) recently reported that Hb ingestion by monocytes drives a novel macrophage phenotype referred to as HA-mac; however, the underlying mechanisms supporting this finding and its functional significance remain unknown.
In the present study, we examined selective macrophage differentiation occurring in areas of intraplaque hemorrhage in human atherosclerosis relative to macrophage-derived foam cells, which contribute to necrotic core formation.
Human tissue specimens
Human plaques were selected from the CVPath Institute Sudden Coronary Death registry (8). All plaques were identified according to a modified American Heart Association classification (9) (see the Online Appendix for details).
Cell culture, flow cytometry, and quantitative polymerase chain reaction
Human monocytes collected from healthy volunteers (Astarte Biologics, Redmond, Washington) were differentiated over 7 days into macrophages in Gibco RPMI 1640 medium (Invitrogen, Carlsbad, California) supplemented with 10% human serum (Invitrogen) (see the Online Appendix for details).
The data are expressed as mean ± SD. A p value <0.05 was considered significant (see the Online Appendix for more details).
M2 macrophage markers are present in human atherosclerotic plaques at regions of previous hemorrhage/angiogenesis/iron and are characterized by expression of MR (CD206) and CD163
To determine the role of hemorrhage in macrophage differentiation, we stained coronary plaques (frozen sections) from individuals who had died suddenly and were referred to our institution (8). We compared staining intensity in regions of previous hemorrhage/angiogenesis/iron (identified by glycophorin A/CD31/iron deposition) to perinecrotic core areas devoid of these markers (Ctrl). As expected, hemorrhagic regions demonstrated significantly greater staining for glycophorin A, CD31, and ferric iron (as identified by Perl Prussian blue stain) compared with Ctrl (Fig. 1A). Although CD68 density was not different, markers of M2 macrophages, mannose receptor (MR) and CD163, were significantly more prevalent in areas of previous hemorrhage/angiogenesis/iron (Figs. 1 and 2)⇓. In Ctrl regions devoid of hemorrhage/angiogenesis/iron, a minority of macrophages stained for these markers. Macrophages in areas of hemorrhage/angiogenesis/iron also demonstrated minimal staining for the proinflammatory M1 markers tumor necrosis factor α (Figs. 1H, 1P, 1S, and 2I) and inducible nitric oxide synthase (Online Fig. 1), which were heavily expressed by foam cells.
By immunostaining (data not shown) and quantitative polymerase chain reaction (QPCR) of RNA extracted from the intima of atherosclerotic plaques from cross-sectional regions poor (i.e., Ctrl) and rich in hemorrhage/angiogenesis/iron (i.e., high MR areas), heme oxygenase-1 was significantly up-regulated (QPCR: 1.34 ± 0.6 vs. 3.7 ± 2.3, p = 0.03). (Compared with RNA collected from Ctrl, areas of hemorrhage demonstrated increased staining for MR/CD163 by immunohistochemistry [Online Table 1]).
For clarity, macrophages expressing high MR/CD163 are referred to as M(Hb) (Hb-stimulated macrophages) to distinguish them from the previously described IL-4–induced M2 phenotype.
CD163/MR-positive macrophages, M(Hb), are distinct from foam cells
M(Hb) exhibit a tissue profile distinct from macrophage foam cells (which do not stain for either MR or CD163). Macrophage foam cells stained positively for oil red O (ORO). Surprisingly, M(Hb) demonstrated very little ORO positivity (Figs. 1D, 1L, 1S, and 2G). The scavenger receptor (SR) CD36, responsible for uptake of oxidized low-density lipoprotein (oxLDL), was also minimally expressed in these compared with foam cells (Ctrl) (Figs. 1F, 1N, 1S, and 2H).
Rabbit model of simulated hemorrhage results in increased expression of IL-10 and MR up-regulation in macrophages
To test whether hemorrhage promotes monocyte differentiation into MR-expressing cells, we used an in vivo model of foam cell formation, harvesting macrophages from subcutaneous sponges exposed to saline versus autologous erythrocytes. Sponges implanted for 28 days yielded between 1 and 2 million macrophages that were >98% positive for the macrophage marker RAM11 and focally ORO positive (Fig. 3A).
The 24-h supernatants from macrophages harvested from sponges exposed to autologous erythrocytes resulted in significantly greater levels of IL-10 by enzyme-linked immunosorbent assay compared with Ctrl sponges (Fig. 3B). There was also a significant increase in mean fluorescence intensity (MFI) for MR staining, as determined by flow cytometry, compared with Ctrls (Fig. 3C). To determine which component of the erythrocyte is critical in stimulating MR up-regulation, sponges were exposed to erythrocyte ghosts devoid of Hb and compared with Ctrl. Macrophages from sponges exposed to erythrocytes ghosts demonstrated similar IL-10 levels as well as MR expression compared with Ctrls (Figs. 3D and 3E). These data implicate exposure to Hb as causal in increased macrophage expression of IL-10 and MR in vivo. To explore whether foam cells could be converted to MR-positive macrophages consistent with recent reports about macrophage plasticity (10), rabbit foam cells were exposed to Hb:Hp (1 mg/ml, a dose previously shown to increase IL-10 expression ) for 1 day. This did not result in any appreciable change in the amount of secreted IL-10 (data not shown) or in MR MFI by flow cytometry (Fig. 3F). Last, rabbit peripheral blood monocytes were collected using anti-CD14 antibody and fluorescent activated cell sorting. Cells were plated and exposed to Hb or Hb:Hp in vitro for 7 days. Exposure of monocytes to Hb alone (1 mg/ml) did not affect IL-10 levels (Fig. 3G) but resulted in an increase in MFI for MR (Fig. 3H), whereas Hb:Hp exposure resulted in a significant increase in both IL-10 (Fig. 3I) and an increase in MFI for the MR (Fig. 3J) by flow cytometry. Collectively, these results indicate that Hb:Hp exposure primes monocytes to increase their expression of IL-10 and MR, but that exposure of macrophage foam cells to Hb:Hp does not.
Human monocytes exposed to Hb:Hp express markers consistent with M2 macrophages but are distinct from IL-4–induced M2 macrophages
To test whether Hb induces macrophage differentiation as in human atherosclerotic plaques, human monocytes were exposed to Hb:Hp (1 mg/ml) for 7 days. (A dose of 1 mg/ml was established by conducting a dose-response curve for IL-10 [Online Fig. 2]. This corresponds to a concentration of 6.7 μM, which is well below the >1 M concentration of Hb found in human erythrocytes ). A comparator set of cells were differentiated in the presence of human IL-4 (15 ng/ml) given 4 h after plating or LPS (100 ng/ml) overnight on day 6 to induce M2 and M1 phenotypes, respectively, while Ctrl cells were kept in normal serum. Cells were analyzed by fluorescent activated cell sorting using anti-mannose and anti-CD163 receptor antibodies. As indicated by a change in MFI, LPS treatment reduced expression of MR compared with Ctrl, whereas IL-4 robustly increased MR and to a much lesser extent CD163 expression, consistent with previous reports (12) (Figs. 4A and 4B). Hb:Hp robustly increased expression of both receptors (Figs. 4A and 4B). Exposure to Hb alone increased expression of both mannose and CD163 receptors but not to the same extent as Hb:Hp (n = 3/group [data not shown]). This is consistent with known low-affinity binding of Hb to CD163 (13). Dual immunostaining of in vitro Hb:Hp differentiated human monocytes demonstrated a distinct population of cells, which expressed increased levels of both receptors (Fig. 4C) in a select population of macrophages by flow cytometry. This was in contrast to IL-4, which up-regulated MR expression in a distinct population of cells.
To confirm the relevance of our findings, we performed dual immunofluorescent staining in 7 atherosclerotic plaques with evidence of hemorrhage. As shown in the representative confocal images in Figure 4D, areas of hemorrhage demonstrated a population of CD163 and CD206 (MR)–positive macrophages, which colocalized with each other. To further verify whether our findings had applicability to human plaques, RNA was extracted from the intima of fibroatheromas with hemorrhage/angiogenesis/iron as described previously. Expression of both MR and CD163 transcripts by QPCR was significantly higher in sections staining for MR compared with those with low MR staining (i.e., Ctrl) (Fig. 4E). A significant correlation between MR and CD163 transcripts was found by multiple linear regression (Fig. 4F). Collectively, these results suggest that differentiating monocytes in the presence of Hb:Hp can reproduce the M(Hb) cell phenotype found in areas of plaque with previous hemorrhage/angiogenesis/iron.
Next, we compared cytokine release from Hb:Hp differentiated macrophages versus those treated with IL-4 or LPS using markers of alternative and classic activation. Culture media from Hb:Hp differentiated macrophages expressed high levels of both IL-10 and IL-1 receptor antagonist similar to macrophages differentiated in IL-4 (Fig. 5A). Hb:Hp differentiated macrophages expressed significantly higher levels of MCP-1 versus Ctrl and IL-4 differentiated macrophages but significantly less than LPS treated macrophages. Similar expression of the same transcripts was seen in RNA generated from human plaques (Online Table 1, Online Fig. 3).
Differentiation of monocytes in Hb:Hp prevents foam cell formation through down-regulation of SRs and up-regulation of ATP-binding cassette (ABC) transporters
To examine the effects of Hb:Hp exposure on macrophage lipid uptake, human monocytes were differentiated as previously described. Seven-day Ctrl, IL-4, and Hb:Hp differentiated macrophages were loaded with oxLDL (30 μg/ml) for 48 h. M(Hb) loaded with oxLDL demonstrated minimal ORO-positive cells compared with both IL-4 and Ctrl cells (Figs. 5B and 5C). A similar experiment using both Ctrl and Hb:Hp differentiated macrophages was conducted and cholesterol accumulation quantitated by enzymatic determination. M(Hb) demonstrated significantly lower levels of total and free cholesterol versus Ctrl cells and were significantly more resistant to oxLDL loading (Fig. 5D). Similar results were obtained using acetylated LDL (50 μg/ml) (data not shown). Although fold increases in total and free cholesterol were greater in M(Hb) than in Ctrl cells, this was not the case for esterified cholesterol in which fold increases were lower (i.e., ≥50%) compared with Ctrl cells. These data indicate that although some uptake mechanisms still seem to be active in M(Hb), those involving production of esterified cholesterol are not as active in M(Hb).
We examined transcription by QPCR of SRs responsible for the uptake of oxLDL in Hb:Hp differentiated versus Ctrl macrophages. M(Hb) demonstrated down-regulation of both class A (I and II) and class B SRs compared with Ctrl (Table 1). Macrophage cholesterol efflux protects cells from free cholesterol– and oxysterol-induced toxicity and is mediated via the ABC transporters ABCA1 and ABCG1 (14). Both were significantly up-regulated in M(Hb) compared with Ctrl (Table 1). Liver X receptor alpha (LXRα) is involved in cholesterol homeostasis through regulation ABC transporters. LXRα was unchanged compared with Ctrl (Table 1).
To confirm our observation that monocytes differentiated in the presence of Hb:Hp up-regulated ABC transporters, we examined the correlation between MR and CD163 receptor and ABCA1 and ABCG1 transcripts by QPCR in human plaques. There was a highly significant correlation between MR/CD163 and both ABC transporter transcripts (Online Figs. 4A and 4B). Given that other cell types also present in the intima are capable of expressing ABC transporters, an approach using laser capture microdissection was performed to isolate macrophage-rich areas. In 6 separate human atherosclerotic plaques with evidence of hemorrhage, laser capture microdissection confirmed up-regulation of ABC transporters in areas rich in CD163/CD206-positive macrophages (CD68) versus macrophage-rich areas containing minimal expression of these receptors (Online Fig. 4C). Further confirmation of changes in ABCA1 (as well as CD36) expression between M(Hb) and Ctrl areas was performed in 5 atherosclerotic plaques with evidence of hemorrhage using immunohistochemistry (Online Fig. 4D).
Intracellular iron affects ABC transporter expression and cholesterol efflux in M(Hb)
Because Hb:Hp treatment increases macrophage gene expression of ABCA1 and ABCG1, suggesting a potential consequence on cholesterol efflux, functional experiments were performed to determine the influence of Hp:Hp on cholesterol efflux to apolipoprotein A-I (apo A-I). We incubated human monocyte/macrophages differentiated in Hb:Hp or Ctrl media with oxLDL (30 μg/ml) for 48 h to induce cholesterol ester accumulation. We subsequently exposed them to apo A-I (100 μg/ml) for 24 h to induce cholesterol efflux and then determined cholesterol levels by enzymatic assay. apo A-I treatment significantly induced cholesterol efflux in M(Hb), as indicated by significantly decreased levels of total, free, and esterified cholesterol levels compared with Ctrl cells (Fig. 6A).
Because iron loading may produce oxidant stress and subsequent modulation of gene expression programs (15), we examined intracellular iron levels in M(Hb). Using calcein fluorescence, an assay based on quenching calcein fluorescence by free iron (16), we found higher levels of calcein fluorescence in M(Hb) compared with Ctrl, consistent with lower intracellular iron levels (Fig. 6B). Ferroportin (FPN), the only known iron exporter in human cells, was increased in M(Hb) compared with Ctrl as examined by Western blotting (Fig. 6C). These results are consistent with previous data, which suggest that Hb loading of macrophages increases expression of FPN facilitating cellular export of iron and thus lowering intracellular iron (17).
To examine whether low intracellular iron is causal in the changes in ABCA1 gene expression, we treated M(Hb) overnight (day 6) with hepcidin, a secreted hepatocyte peptide that controls iron homeostasis by negatively regulating expression of FPN and thus increasing intracellular iron (18). (The efficiency of hepcidin in increasing intracellular iron and degrading FPN was confirmed in M(Hb) by calcein fluorescence and Western blot, respectively [Online Figs. 5A and 5B]). Overnight treatment of M(Hb) with hepcidin (700 nM) significantly decreased expression of ABCA1 by QPCR (Fig. 6D) but had no effect on its expression by Ctrl macrophages. Next, functional experiments were performed to determine the influence of increasing intracellular iron (using hepcidin) on cholesterol efflux to apo A-I in M(Hb). M(Hb) cells were treated overnight with hepcidin after 48 h of cholesterol loading and subsequently exposed to apo A-I to induce cholesterol efflux. Consistent with our earlier results, M(Hb) demonstrated decreases in total, free, and esterified cholesterol when exposed to apo A-I. These effects were completely abrogated by hepcidin treatment (Fig. 6E). These data suggest a critical role for intracellular iron in modulating ABC transporter expression and cholesterol efflux activity in M(Hb).
Reduction of reactive oxygen species mediates LXRα activation of ABC transporters in M(Hb)
Given the important role of iron in generating reactive oxygen species (ROS), we next examined H2O2 production in Ctrl and M(Hb) cells with and without overnight hepcidin treatment. H2O2, as measured by 24-h Amplex red assay (Invitrogen), was significantly reduced in M(Hb) compared with Ctrl cells, consistent with previous reports about the antioxidant response of these cells to Hb:Hp loading (7) (Fig. 7A). Overnight hepcidin treatment of M(Hb) cells resulted in a significant increase in H2O2 production (similar to the level seen in Ctrl cells) (Fig. 7A). These data demonstrate that iron efflux plays an important role in ROS modulation in M(Hb).
To characterize ROS generation, especially the Fenton reaction product OH° in M(Hb) in vivo, we used the redox-sensitive dye dihydrorhodamine-123 (DHR) on cryosections of freshly isolated human carotid atherosclerotic lesions that contained M(Hb) and foam cells. M(Hb) areas demonstrated dull green fluorescence that was not nearly as intense as seen in foam cells, suggesting relatively reduced ROS generation (Online Fig. 6).
H2O2 was the primary responsible for the DHR signal in M(Hb) as co-incubation of the sections with the H2O2 scavenger catalase quenched most of the fluorescence activity. Co-incubation with dimethyl sulfoxide, the scavenger for OH°, only very mildly attenuated the signal (Online Fig. 6). Identical results were found in all carotid plaques studied. Collectively, these results suggest that in M(Hb), lower levels of free intracellular iron (Fe2+) results in less production of the highly toxic Fenton product OH°.
OH° and other ROS such as peroxynitrite (ONOO°) can severely damage DNA and also lead to protein nitration. We assessed oxidant stress–induced DNA damage in areas of plaques with and without the M(Hb) phenotype using an antibody specific for oxygen-adducted guanosine. As seen in Online Figure 7, although pericore foam cells stained heavily for oxygen-adducted guanosine, M(Hb) demonstrated minimal staining, indicating reduced oxidant stress–induced DNA damage. Similarly, an antibody against 3-nitrotyrosine, a selective marker for protein nitration due to ONOO°, demonstrated minimal staining in M(Hb) areas (Online Fig. 1).
To further investigate whether a decrease in oxidant stress plays an important role in up-regulation of ABC transporters, we differentiated human monocytes in the antioxidant superoxide dismutase (SOD) (75 U/ml) for 7 days. RNA from Ctrl and SOD-treated macrophages was analyzed by QPCR for both ABCA1 and ABCG1. As seen in Figure 7B, SOD treatment significantly increased transcription of both genes. Next, to demonstrate the mechanism by which Hb:Hp treatment increases ABCA1/ABCG1 transcription, genetic knockdown of LXRα, an inducible transcription factor known to be important in human macrophage ABC transporter transcription (14), was conducted. The up-regulation of ABCA1 and ABCG1 transcripts, as detected by QPCR in 7-day Hb:Hp differentiated macrophages, was significantly inhibited by treatment with LXRα small interfering RNA but had no effect on ABCA1/ABCG1 expression in Ctrl-treated cells (Fig. 7C). These data indicate a critical role of LXRα in mediating induction of ABC transporter expression in M(Hb).
Similar experiments were conducted in SOD-treated monocyte/macrophages to examine whether LXRα was also required for up-regulation of ABC transporters in these cells. The up-regulation of ABCA1 and ABCG1 transcripts, as detected by QPCR in 7-day SOD differentiated macrophages, was significantly inhibited by treatment with LXRα small interfering RNA. These data suggest that in SOD-treated macrophages, LXRα is required for up-regulation of ABC transporter expression.
Next, we examined LXRα transcriptional activity in Ctrl-, SOD-, and Hb:Hp-treated monocyte/macrophages by transfecting differentiating cells with a lentivirus encoding an inducible LXRα-responsive firefly luciferase reporter. Luciferase activity at 96 h after transfection was significantly higher in both Hb:Hp- and SOD-treated monocyte/macrophages compared with Ctrl (Fig. 7E). To demonstrate a critical role for intracellular iron in induction of LXRα activity, we treated Hb:Hp differentiating cells transfected with the LXRα reporter as described earlier. Cells were then treated with hepcidin (700 nM) 18 h before harvest to prevent iron efflux and thus increase ROS. This resulted in significant reduction in luciferase activity compared with M(Hb) cells not treated with hepcidin (Fig. 7E). Collectively, these data indicate an important role for intracellular free iron and ROS in activation of LXRα-mediated transcription of ABC transporters in human macrophages.
Recently, it was recognized that different populations of macrophages reside within human plaques and that the microenvironment plays a pivotal role in their differentiation programs (1). Our work significantly expands and integrates previous observations to demonstrate that M2 macrophages first described by Bouhlel et al. (1) in carotid atherosclerotic plaques are identified in areas of hemorrhage/angiogenesis/iron, are distinct from IL-4–induced M2, and result from Hb ingestion by monocytes (Fig. 8). These cells were first reported by Boyle et al. (7) in areas of hemorrhage but are now further characterized by us as having high surface expression of M2 markers MR and CD163. We choose to use the term M(Hb) to refer to this subset instead of the previously used terms M2 or HA-Mac because this subset is induced by Hb not IL-4 or hemorrhage (i.e., not platelet or other white cell derived), respectively.
Importantly, our data also expand on previous observations to demonstrate conclusively the nonfoamy nature of these cells and reveal the mechanisms by which modulation of intracellular iron and ROS regulate cholesterol efflux proteins. Consistent with data from human plaques, M(Hb) were resistant to foam cell formation through down-regulation of SRs and up-regulation of ABC transporters by in vitro assays. Hb loading of macrophages increased ABC transporter expression and cholesterol efflux to apo A-I by reducing intracellular iron and ROS generation through up-regulation of FPN. These effects were reversed by increasing intracellular iron levels with hepcidin and emphasize the importance of iron homeostasis in directing cholesterol handling by macrophages.
Bouhlel et al. (1) were the first to report that cells within atherosclerotic carotid plaques demonstrate expression of M2 markers. Although IL-4 has been classically described as the prototypical stimulus for such conversion, it remains controversial whether such a paradigm exists in human atherosclerosis. Recently, Chinetti-Gbaguidi et al. (2) described the presence of CD68+MR+ macrophages in human plaques that contained smaller lipid droplets. Although this is consistent with our findings, the authors attribute the existence of these cells to IL-4–induced differentiation. In contrast to our findings, they report decreased levels of ABCA1 by QPCR and immunostaining in human plaques compared with CD68+MR− cells, suggesting lower cholesterol efflux capacity. It should first be noted that these data contradict the findings of others who reported increased ABCA1 expression in IL-4–differentiated M2 mouse macrophages, although it remains unclear to what extent these data can be applied to the behavior of human macrophages (19). Although it is beyond the scope of this paper to discuss every discrepancy that exists with our findings, one major weakness that deserves attention is the lack of clear identification of the source of IL-4 responsible for M2 differentiation in human plaques. Previous work found very little expression of IL-4 in human plaques (20). Here we demonstrate conclusively that the locations at which CD68+MR+macrophages are found are in areas of previous hemorrhage. Moreover, we were able to reproduce using Hb:Hp both in an animal model and in vitro, not only the surface markers (MR and CD163) that characterize this macrophage subtype but also the profile of cytokine release as well as the mechanism by which they resist lipid deposition. Human macrophages expressing these surface markers could only be generated by exposure of human monocytes to Hb:Hp and not by IL-4. Although lipid loading as assessed by ORO staining was attenuated by IL-4 differentiation, the effect of Hb:Hp was more striking with regard to this effect. Our data are further supported by the observations of Boyle et al. (7) who also reported that macrophages in areas of hemorrhage are characterized by surface expression of CD163 and display a nonfoam cell phenotype. Collectively, these data emphasize the importance of Hb in macrophage differentiation programs within atherosclerotic plaques.
Although it has long been suspected that iron depletion might play a favorable role in risk of cardiovascular disease (21), whether such an effect really exists and the mechanisms underlying it remain uncertain. Although our data cannot answer this question completely, they do support a role for how alterations in macrophage iron levels are important in controlling cholesterol efflux proteins. We demonstrate that lowering of intracellular iron within M(Hb) is causal in increasing both ABCA1 and ABCG1 expression through its effect on lowering ROS. Knockdown of LXRα in M(Hb) significantly reduced the effects of Hb:Hp on up-regulation of ABCA1 and ABCG1, suggesting that these effects are mediated in part by LXRα transcriptional activation of these genes. These findings were further confirmed by luciferase reporter assays. The exact mechanism by which lowering ROS activates LXRα transcription deserves further attention and is beyond the scope of the present study.
We would be remiss if we did not discuss briefly the limitations of this study. Autopsy examination involves certain selection bias in that these findings may not be representative of humans with intraplaque hemorrhage who survive this event. It is possible that the selection of plaque types may have biased our findings, but this is unlikely because we found a similar phenotype of macrophages in all atheromas with evidence of previous hemorrhage/angiogenesis/iron.
Collectively, these data may help us understand better macrophage diversity within human atherosclerosis. Moreover, our findings suggest that therapies aimed at altering iron homeostasis within macrophages may represent a new avenue to prevent foam cell formation and atherosclerotic lesion progression.
The authors thank the Emory Core Flow Cytometry Lab and the Emory Laser Capture Core for their help. They also thank Lila Adams, Deborah Howard (CVPath), and Michael John (Emory) for their technical support.
For an expanded Methods section and supplemental tables and figures, please see the online version of this article.
This study was supported by the Carlyle Fraser Heart Center, CVPath Inc. and the National Institutes of Health grant RO1 HL096970-01A. All authors have reported that they have no relationships relevant to the contents of this paper to disclose. Edward A. Fisher, MD, served as Guest Editor for this paper.
- Abbreviations and Acronyms
- ATP-binding cassette
- apo A-I
- apolipoprotein A-I
- hemoglobin:haptoglobin complex
- liver X receptor alpha
- mean fluorescence intensity
- hemoglobin-stimulated macrophages
- mannose receptor
- oil red O
- oxidized low-density lipoprotein
- quantitative polymerase chain reaction
- reactive oxygen species
- superoxide dismutase
- scavenger receptor
- Received August 16, 2011.
- Revision received October 5, 2011.
- Accepted October 11, 2011.
- American College of Cardiology Foundation
- Chinetti-Gbaguidi G.,
- Baron M.,
- Bouhlel M.A.,
- et al.
- Nagy E.,
- Eaton J.W.,
- Jeney V.,
- et al.
- Philippidis P.,
- Mason J.C.,
- Evans B.J.,
- et al.
- Kramer M.C.,
- Rittersma S.Z.,
- de Winter R.J.,
- et al.
- Virmani R.,
- Kolodgie F.D.,
- Burke A.P.,
- Farb A.,
- Schwartz S.M.
- Watanabe J.,
- Grijalva V.,
- Hama S.,
- et al.
- Schaer D.J.,
- Schaer C.A.,
- Buehler P.W.,
- et al.
- Yvan-Charvet L.,
- Wang N.,
- Tall A.R.
- Marcil V.,
- Delvin E.,
- Sane A.T.,
- Tremblay A.,
- Levy E.
- Abboud S.,
- Haile D.J.
- Nemeth E.,
- Tuttle M.S.,
- Powelson J.,
- et al.