Introduction

Activation of various cell types contributes to the advancement of atherosclerotic changes in arteries. These cells include endothelial cells, smooth muscle cells, lymphocytes, monocytes, and macrophages [1]. While all of these cell types contribute to the formation and progression of atherosclerosis, macrophage retention within the arterial wall is the sine qua non of atherosclerosis. Macrophages are the major inflammatory cells involved in its progression. Prevailing paradigms describe atherosclerosis as a T-helper type 1 (Th1)-driven disease in which the recruitment and retention of monocytes and lymphocytes in the arterial wall result from an increase in circulating low-density lipoprotein (LDL)-cholesterol in the blood and the accumulation of oxidized LDL in the subendothelial space. Macrophages scavenge lipoprotein particles, transforming into foam cells which secrete pro-inflammatory molecules and play an important role in collagen and matrix breakdown leading to plaque rupture [2,3,4].

Apoptosis of these resident macrophages in the lipid core of the lesion leads to progression of atherosclerosis. Phagocytes, mainly macrophages, play a role in the clearance of apoptotic cells. These apoptotic cells are recognized as dead cells in the process of programmed cell removal or ‘efferocytosis’ [5]. In early lesions, macrophages have a capability of clearing apoptotic cells to avoid progression of atherosclerosis. However, in progressive lesions, macrophages are no longer able to engulf dead cells and eventually the necrotic core is formed and expanded by the accumulation of apoptotic debris. Necrotic core expansion is strongly associated with inflammation, risk for plaque rupture, and thrombus formation [5]. Throughout the body, macrophages play an important role in maintaining efferocytosis, ensuring the rapid clearance of dead cells to prevent inflammatory consequence-associated apoptotic debris. Impairment of these mechanisms in atherosclerotic lesions appears to be critical in the progression of plaques [6].

Recently, it was reported that macrophages themselves can proliferate within atherosclerotic lesions. An increase in macrophages was observed in mouse lungs without any interaction of circulating monocytes, an observation which challenged existing beliefs that tissue-resident macrophages were incapable of proliferating [7]. Both resident and recruited macrophages were thought to be mediated through the action of IL-4 which was sufficient to drive accumulation of macrophages through self-renewal [8]. Additionally, although macrophages in early lesions are derived from recruited monocytes, proliferation of macrophages occurs in advanced mouse atherosclerosis and microenvironmental signals drive their proliferation [9]. Therefore, the capability of local proliferation may be related to differentiation subtypes of resident macrophages [7, 10].

While the role of lipids and cell death dominated the field of atherosclerosis for many years, only more recently have we come to appreciate the heterogenous nature of macrophages. Microenvironment directs monocytes in functionally distinct cell types described within the classic binary paradigm described by Gordon [11]. Within this context, at least two different subtypes of macrophages so-called M1 and M2 macrophages were initially described. Activated (M1) macrophages are induced by a type 1 T-helper cell (Th1) cytokines such as interferon and lipopolysaccharide and are involved with pro-inflammatory activities. Alternative M2 macrophages are stimulated by Th2 cytokines such as cytokine interleukin (IL)-4 or IL-13 and produce anti-inflammatory cytokines such as IL-10 [12] and counterbalance inflammatory responses to M1 macrophages by promoting the resolution of inflammation and inducing tissue repair. Although M1/M2 dichotomy provides a conceptual framework for our understanding of the function of macrophages in the setting of injury, the mechanisms by which macrophages orchestrate inflammation and its resolution to promote tissue repair are incompletely understood.

Evidence for the existence of M2 macrophages in human atherosclerosis was shown by Chinetti-Gbaguidi et al. using histological analysis of human atherosclerotic plaques [8]. M2 macrophages were characterized as a subpopulation of CD68 cells which were positive for the mannose receptor, a surface marker for M2 macrophages. This and many other investigations have helped us go beyond the initial belief that atherosclerosis is primarily driven by Th1-type inflammatory responses [13, 14]. Because macrophages differentiate into morphologically and functionally distinct phenotypes depending upon their microenvironment, the differentiation of macrophages within atherosclerotic plaques is related to cellular events that take place there which may be dependent upon the stage of development of atherosclerotic lesions. Distinct macrophage subtypes can be identified based on their differential expression of surface markers and chemokine receptors [11, 15]. The prevailing view is that M1 macrophages are enriched in progressing lesions while M2 macrophages are present in regressing lesions but even this understanding is beginning to change.

Recent studies have demonstrated additional complexity to this paradigm. In general, the M2 phenotype has been divided into subgroups depending upon their activation stimuli with M2a macrophages induced by IL-4 or IL-13 and M2b induced by IL-1β or LPS. However, other subtypes have also been described such as M4 macrophages induced by CXCL4, Mox macrophages induced by oxidized LDL, and M (Hb) or Mhem macrophages induced by hemoglobin in the setting of intraplaque hemorrhage. As mentioned before, each has a distinctive role in plaque evolution and our understanding of their various roles with respect to atherosclerosis is still evolving and to this day is not completely understood. As an example, studies using single-cell RNA-sequencing methods revealed new clusters of macrophages by transcriptome signatures at single-cell resolution and suggested that the foamy macrophages are not related to inflammatory reactions [16,17,18].

In this review, we summarize and focus on the differential subtypes of macrophages in atherosclerotic plaques and their functional roles within different milieu such as lipid, intraplaque hemorrhage (IPH), and plaque regression.

Distinct types of macrophage in atherosclerosis

Monocytes enter into the plaque through processes such as capture, rolling, and transmigration with each step regulated by multiple factors. Lipoprotein, especially in its modified forms, accumulates in the proteoglycan rich layer of the intima where it is ingested by macrophages giving rise to foam cells.

Activation of these cells by toll-like receptors (TLR) as well as interferon-gamma induced by engulfed lipoproteins polarizes them into M1 macrophages. Pro-atherosclerotic cytokines such as IL-6, IL-1b, tumor necrosis factor (TNF), IL-23 and IL-12 are derived from M1 macrophages which sustain the inflammatory response [19]. M1 macrophage plays a role in killing intracellular pathogens and their polarization to this phenotype also has other consequences within the context of atherosclerosis. Reactive oxygen and nitrogen species are generated by M1 macrophages to eliminate bacterial, fungal, and viral infections, which also may worsen oxidative stress in the plaque [20]. Further M1 macrophages expressing different chemokine (C-X-C motif) receptor ligands (i.e., CXCL9, CXCL10, and CXCL5) induce the recruitment of Th1 and natural killer cells to kill intracellular pathogens [21]. These functions have important benefits especially in acute infection but they cause tissue damage and impair wound healing under the milieu of sterile inflammation generated within the context of lipid-driven atherosclerosis [22]. In atherosclerotic lesions of humans, M1 macrophages are located in lipid-rich areas away from M2 macrophages [8]. These cells have decreased ability to migrate, and their accumulation and death are responsible for necrotic core formation and enlargement, a key feature of advancing plaques.

In in vitro and mouse studies, M1 macrophages were distinguished from M2 macrophages. The former induce plaque inflammation and the latter are thought to be involved with the resolution of inflammation and tissue repair. M2 macrophages are induced by IL-4 and IL-13 cytokines which are secreted by Th2 cells [12, 23]. Therefore, the term M2 macrophage is influenced by the TH2 nomenclature.

M2 macrophages have been classified into three different subtypes including M2a, M2b, and M2c macrophages. IL-4 and IL-13 trigger M2a macrophages which express high levels of the mannose receptor [24,25,26,27,28]. M2a macrophages secrete pro-fibrotic factors (i.e., fibronectin, insulin-like growth factor (IGF), and transforming growth factor β (TGF-β) which contributes to the tissue repair [24,25,26,27,28]. M2b macrophages are triggered by exposure to immune complexes and TLR ligands or IL-1 receptor agonists [29]. These cells produce not only anti-inflammatory (IL-10) and pro-inflammatory (IL-1, IL-6, TNFα) cytokines but also TGF-β1. M2c macrophages are triggered by IL-10 and glucocorticoids. M2a macrophages are classically recognized as “wound healing macrophages”, whereas M2b and M2c macrophages are called as “regulatory macrophages” [29]. All M2 macrophages have an anti-inflammatory cytokine profile such as low production of IL-12 and high production of both IL-10 and TGF-ß [21]. However, the phenotypic spectrum of macrophages in vivo is not as simple as this schema might suggest. Thus, it remains unclear how this above subclassification of M2 macrophages relates to the specific roles on different subtypes of M2 macrophages in atherosclerosis.

More recent data suggest a role for M2 macrophages in atherosclerosis regression in mouse models [30, 31]. The fact that M2 macrophages have been found in human plaques suggests there is potential for these cells to cause plaque regression but role they actually play in this process in humans is unknown. Common to experimental models of atherosclerosis (mostly mouse studies) is the findings that during regression plaque macrophages decrease and in some cases become enriched in M2 macrophages [30, 32]. But these results cannot directly be translated into humans because macrophage subtypes detected in mice and humans do not directly coincide, and mouse and human atherosclerotic plaque evolution are very different.

This classification can be further broadened into other types of macrophages that have been found in either mouse or human atherosclerosis or both [33] (Fig. 1). Mox macrophages have been found in mouse atherosclerosis where they are induced by oxidized LDL and have proatherogenic properties. M4 macrophages are induced by platelet chemokine CXCL4 and are identified by loss of the hemoglobin–haptoglobin scavenger receptor CD163 and express a combination of CD68, S100A8 and MMP7. They are predominated expressed in the adventitia and intima and have been associated with plaque instability [34]. Advanced plaques also may develop intraplaque hemorrhage (IPH) in which the release of free Hb and its subsequent uptake binding by haptoglobin via the CD163 receptor causes differentiation into M (Hb) or Mhem macrophages. As will be discussed below, emerging data both in human and mouse studies suggest these macrophages may be pathogenic in the setting of IPH.

Fig. 1
figure 1

Main macrophage subtypes found in atherosclerotic lesions. Stimuli present in atherosclerotic lesions drive the differentiation of monocytes towards different macrophage phenotypes. a M1 macrophages release pro-inflammatory cytokines. b M (Hb), Mhem, and M2 macrophages are resistant to lipid accumulation, possess iron-handling capacities, and have anti-inflammatory effects. c Mox macrophages display an antioxidant gene expression profile. d M4 macrophages, like M1 macrophages, are pro-inflammatory, but lack the capacity for phagocytosis. COX-2 cyclooxygenase, CXCL4 C-X-C motif chemokine 4, HMOX-1 haem oxygenase (decycling) 1, LDL low-density lipoprotein, LXR liver X receptor, MMP-7 matrix metalloproteinase-7, NFE2L2 nuclear factor (erythroid-derived 2)-like 2, NF-κB nuclear factor kappa-light-chain-enhancer of activated B cells, TLR toll-like receptor, TNF tumor necrosis factor (Ref. [33])

Full size image

M2 macrophages were first identified in human carotid endarterectomy lesions by Bouhlel et al. Two distinct types of cells were described: (1) M1 macrophages characterized by activation of MCP-1, IL-6, and TNF-alpha; and M2 cells with upregulation of MR (CD206) and CD163, CCL-18, and IL-10. Staining for MR and MCP-1 demonstrated distinct and separate tissue localizations. A correlation was shown between PPAR gamma mRNA levels and markers of M2 but not M1 macrophages [14]. In vitro priming of human monocytes increased expression of MR but not CD163. Although the authors stated these data showed the presence of M2 cells in advanced atherosclerotic lesions, the stimulus for M2 differentiation was not clearly resolved, as there is very little IL-4 expression that has been found in human atherosclerotic plaques [35]. Since the same cells that were described by Bouhlel were also found by the same group to contain very little in the way of lipid droplets, it remains uncertain to what extent these cells overlap with hemoglobin-induced macrophages described below [8].

Because different subtypes of M2 macrophages express some of the same markers this suggests the presence of the significant overlap which might make their phenotypic differentiation difficult. From this point of view, macrophage phenotypes should be categorized considering not only the expression of surface markers but also their specific function. Notably, it is believed that macrophages are not terminally differentiated but can switch from one phenotype to another through changes in microenvironment [36, 37]. The diversity of macrophage phenotypes within atherosclerosis has become a hot topic and still remains controversial.

What single-cell RNA-seq studies tell us about macrophage populations in atherosclerosis

The characterization of macrophages subtypes in atherosclerosis has been challenging and continues to evolve as our techniques to evaluate them continue to change. With the recent advent of single-cell RNA-seq (sc-RNA-seq) methods, new clusters of macrophages have been described in atherosclerosis by transcriptome signatures at single-cell resolution.

Two sc-RNA-seq studies in Ldlr−/− [10] and Apoe−/− mice [18] used an unsupervised clustering method to map leukocyte heterogeneity. Both showed the macrophages and T cells are the predominant leukocytes in atherosclerotic plaques, and the proportion of macrophages increases as the plaques progress. Different clusters of macrophages have been described, the inflammatory cytokines (e.g., IL-1β, Cxcl2, Ccl2, and TNF were mostly expressed in non-foamy “inflammatory” macrophages [17, 18]. Foamy macrophages expressed high Trem2, Cd9, Ctsd, or Spp1 (osteopontin), but express lower levels of inflammatory cytokines [16]. Using single-cell RNA sequencing and genetic fate mapping, a recent study showed profiling of CX3CR1+ precursors cells during both progression and regression of atherosclerosis revealed a spectrum of macrophage activation states. This includes “resident-like macrophages”, “inflammatory macrophages”, as well as “TREM2 hi macrophages” (Folr2 hi, chemokine hi, and Trem2 hi macrophage, respectively). Differentiated and activated macrophages were highly enriched in cells from progressing versus regressing plaques. Interestingly, a cluster of CX3CR1 high expressing macrophages, distinct from other macrophage populations, maintain a stem cell-like proliferative state, are enriched for cell cycle genes, and have been termed “stem-like macrophages”. These results indicate that plaque microenvironment determines stimulating factors that could drive the activation and differentiation of macrophages as plaques progress. Reversal of hyperlipidemia improves the microenvironment in the plaque during regression, reducing the heterogeneity and complexity of macrophages [38]. These single-cell RNA-seq studies provided important transcriptome data in lesional cell populations, and by combining the lineage tracing models as well as proteomic profiling, more studies will further reveal the diversity of macrophage phenotypes and their functions in disease pathogenesis. However, still missing in such experimental models is the link to human atherosclerosis where such dynamics are much less understood. More recently, Fernandez et al. conducted single-cell proteomic and transcriptomic analyses of human carotid plaques of patients with symptomatic and asymptomatic disease. Plaques from symptomatic patients contained alternatively activated macrophages, including CD163 positive subsets. Distinct macrophage subsets displayed activated and pro-inflammatory functional states, genes expressed in lipid metabolism, and unique anti-inflammatory signatures, suggesting that subsets of macrophages within the human plaque environment are not accurately defined by M1 and M2 phenotypical descriptions characterized by certain surface markers [39].

Efferocytosis

Efferocytosis is one of the most basic functions of macrophages. Phagocytic clearance of apoptotic macrophages in early lesions is usually efficient, giving physiological benefit by tempering inflammatory responses. That is because areas with early lesional apoptotic macrophages are cleared before they become necrotic, with release of anti-inflammatory cytokines and pro-resolving lipids. On the other hand, apoptotic macrophages are more frequently observed within advanced lesion relative to early lesions. This fact suggests that phagocytic clearance in advanced lesions is not effective. In human carotid arteries, it has been shown that a fair amount of apoptotic macrophages were not engulfed by phagocytes. On the other hand, in human tonsillar tissue, apoptotic macrophages were mostly found inside phagocytes. Many studies demonstrated that areas close to the necrotic core have larger concentrations of lesional apoptotic macrophages [40,41,42,43]. Defective clearance of these cells leads to macrophage necrosis and enhanced inflammatory reactions as well as release of immunogenic antigens [44,45,46]. In fact, in the Apoe−/− mouse model of atherosclerosis defective clearance of apoptotic cells correlated with an increase in the level of inflammatory markers including TNF-α [47]. Eventually, defective phagocytic clearance of apoptotic macrophages provokes plaque progression through several important mechanisms. First, non-cleared apoptotic cells provide a source for pro-coagulant molecules, leading to thrombus formation [48]. Second, the necrotic core consisting of apoptotic and necrotic cells leads to increases in lysophosphatidic acid derivatives which are associated with platelet aggregation and prevention of emigration of macrophages out of the lesion [49, 50]. Third, macrophages secrete matrix-degrading proteases via inflammatory factors in the necrotic core [51]. Lastly, antigenic factors such as thymidine phosphorylate are released from necrotic macrophages, accelerating atherosclerotic changes such as intraplaque angiogenesis and inflammation [52, 53].

Although it is not fully understood why the inefficiency of phagocytic clearance of apoptotic cells is observed within advanced lesions, previous studies have introduced important mechanisms [54]. One such mechanism is the role of oxidized molecules which inhibit the uptake of apoptotic cells by phagocytes [54]. Competitive inhibition of receptors occurs since the oxidized molecules are recognized by certain phagocyte receptors expressed on apoptotic cells. Also, oxidized lipid may affect the affinity of phagocytes for detecting and taking up apoptotic cells [55, 56]. Recognition and ingestion by phagocytes are inhibited because oxidized LDL competes with apoptotic cells for macrophage binding, suggesting they both share oxidatively modified moieties on their surface than serve as ligands to macrophage recognition [57]. However, clinical trials in humans have been able to show that antioxidants were able to prevent progression of atherosclerosis. Perhaps the most powerful evidence for the role of efferocytosis in plaque progression was found by demonstrating that an antibody to CD47, a key ‘don’t eat me’ molecule reversed defective efferocytosis, normalized clearance of diseased vascular tissue and reduced atherosclerosis in multiple mouse models [58].

Relationship between macrophages and lipid

Oxidative stress-modified LDL generates signals that are recognized by pattern recognition receptors (PRR) on macrophages and other immune cells [4]. Scavenger receptors are a type of PRR expressed by macrophages and numerous ones have been identified such as scavenger receptor A1, scavenger receptor B1, CD36, scavenger receptor B1, etc., all of which can bind oxidized LDL and promote foam cell formation. In the late endolysosomal compartment, cholesterol esters from lipoprotein are hydrolyzed to free cholesterol and fatty acids. In vitro SRA1 and CD36 mediate > 75% of the degradation of modified LDL with combined deficiency of both receptors reducing but not totally preventing foam cell formation in mice, suggesting other mechanisms of LDL uptake exist in vivo [59, 60]. Other studies suggest native LDL may also be taken into macrophages by pinocytosis which also delivers it to the endolysosomal compartment where it undergoes esterification [61].

The ability of macrophages to store cholesterol can be overwhelmed in the setting of abundant cholesterol uptake and results in pathological changes to the function of these cells. Cholesterol esters are mostly biologically stable whereas free cholesterol can cause significant toxicity to cells. As free cholesterol builds up in cells, the ability to esterify it via acetyl-coenzyme A: cholesterol acetyltransferase 1 (ACAT1) starts to decline which further promotes its accumulation. This enhances inflammatory signaling via TLR signaling and NF-kB activation. Because of impaired trafficking of free cholesterol out of lysosomes cholesterol efflux may also become impaired. As a result of these changes in cholesterol metabolism, endoplasmic reticulum stress results in cell death.

Phagocytic function of macrophages is extremely important in the clearance of dead cells. The process of programmed cell removal (or efferocytosis) is dysregulated in atherosclerosis. Defects in lipid metabolism and upregulation of CD47, an important “don’t eat me” molecule that has been implicated in defective efferocytosis. The combination of lipid toxicity and defective dead cell clearance results in necrosis and the release of cell components and lipids that contribute to the growth of the necrotic core, a key feature of advanced atherosclerotic plaques.

Cholesterol crystals are also present at the early stage of atherosclerotic lesions and stimulate inflammatory activation of macrophages [62]. In lipopolysaccharide (LPS)-primed human peripheral blood mononuclear cells and in vivo in mice, cholesterol crystals are responsible for the stimulation of the caspase-1-activating-NLRP3 inflammasome and which causes the cleavage and the release of IL-1 family cytokines (including IL-1β and IL-18) [62]. Cholesterol crystals accumulate in atherosclerotic lesions and play a role in the differentiation of M1 macrophages. A causal role in the activation of the NRLP3 inflammasome in atherosclerosis progression remains controversial with some mouse models with deficiency in components of the inflammasome showing decreased atherosclerosis while others failing to show a protective effect of its deletion [62, 63].

Most aspects of lipid and its by-products drive the differentiation of macrophages into M1 cells. Oxidized lipoproteins inhibit Kruppel-like factor 2 (KLF2) while minimally modified LDL (mmLDL) acts via activated TLR-4 and TLR-2 pathways [64]. Cholesteryl linoleate, the major cholesteryl ester in atherosclerotic plaques, stimulates M1 polarization via a TLR4/NF-kB dependent mechanism while its oxidation products such as 7-keto-cholesteryl-9 carboxinonanoate promote an anti-inflammatory phenotype dependent TGF-β signaling pathway [65,66,67].

Responses to hemorrhage

More than 75 years ago, Wartman et al. reported that IPH might be responsible for the progression of coronary artery lesions [68]. Neovascularization often occurs in atherosclerotic plaques and can result in IPH after vessel rupture [69]. Barger et al. have shown a rich microvascular network extending from the adventitia through the media and into the thickened intima after the injection of silicone polymer into atherosclerotic human coronary arteries [70]. IPH is the essential trigger for alternative macrophage conversion in the atherosclerotic lesions. IPH is caused by plaque neovascularization, increased microvessel permeability, and leakage of blood [71, 72] (Fig. 2). These findings are recognized as one of features of advanced atherosclerotic plaques, are related to rapid progression of atherosclerotic lesions with accumulation of erythrocyte membranes, and promote necrotic core enlargement [69]. In a large series of human plaques from victims of sudden coronary death, there was a greater frequency of previous IPH (as identified by immunostaining for glycophorin A, an anion exchange protein specific to erythrocytes) in high-risk plaques prone to rupture compared to early lesion morphologies and stable plaques. IPH is thought to result in expansion of the necrotic core through deposition of free cholesterol-rich erythrocyte membranes. As the amount of glycophorin A increased so did macrophage infiltration, suggesting hemorrhage is an inflammatory stimulus.

Fig. 2
figure 2

M2 markers CD206/CD163 are present in human atherosclerotic lesions at sites of prior hemorrhage and are distinct from foam cells. a Representative frozen section of human coronary fibroatheroma from a 48-year-old man who died suddenly. High-power image from ctrl (red box) shows foamy macrophages in the perinecrotic core (NC) region. The blue box shows an area rich in angiogenesis and iron (low-power image, Movat stain; high-power images, hematoxylin and eosin [H&E] stain). Photomicrographs of the boxed areas from ctrl (red boxes) (c–j) and area of angio/hemo/iron (blue boxes) (k–r). Note that the ctrl area shows a lack of CD31 staining (brown). b Plaque regions were identified by the presence of angiogenesis/hemorrhage/iron (angio/hemo/iron) and compared with control (ctrl) pericore regions of macrophages devoid of angio/hemo/iron (c), abundant oil red O (ORO) positivity (red) (d), macrophage infiltration (CD68, brown) (e), CD36 staining (brown) (f) and no iron staining (blue, Perls’ Prussian blue) (g). It also demonstrates abundant tumor necrosis factor (TNF-) positivity (brown) (h), but there is minimal MR (CD206, brown) (i) and CD163 (brown) (j) immunostaining. k–r From an area of angio/hemo/iron showing abundant CD31 staining (k, black arrows point to angiogenesis), rare positive cells for oil red O (l), but abundant C68 staining (m). This area also demonstrates minimal CD36 staining (n), but abundance of iron (o), minimal TNF staining (p), but positive staining for CD206 and CD163 (q, r). Quantitative analysis (s) from ctrl and angio/hemo/iron areas from 14 plaques demonstrated equivalent macrophage area density (CD68) but higher expression of CD163 and CD206 in regions of angio/hemo/iron than in ctrl regions. Scale bars: low-power, 2 mm; high-power: 200 m (non-normal distribution CD68 and CD163). Fe iron (Ref. [71])

Full size image

IPH results in erythrocyte lysis through oxidative stress and afterward free hemoglobin is released (Hb) [73]. Hb is immediately complexed with the plasma protein haptoglobin and macrophages internalize hemoglobin:haptoglobin (HH) complexes via CD 163 receptor which is exclusively expressed on macrophages [74, 75]. CD163 is involved in clearance and endocytosis of HH, and is necessary for effective hemoglobin clearance after IPH [76]. The heme subunit of hemoglobin is degraded by the heme oxygenase enzymes (HO) which produces antioxidants carbon monoxide and biliverdin and also releases free Fe2+. Once iron has been released, it is either used by the cell, sequestered in a non-toxic form as ferritin, or exported from the cell via the iron exporter ferroportin (FPN) which also converts it to the less redox-active form Fe3+ by ceruloplasmin. Although elements of this process such as HO have been studied within the context of atherosclerosis, our understanding of the impact of how IPH affects macrophage function and the process of atherogenesis remains incomplete. Ingestion of HH by CD163 results in differentiation of macrophages into a subtype we have termed M (Hb) and others Mhem. M (Hb) macrophage is mainly located in areas of neoangiogenesis and hemorrhage, and characterized by high surface expression of CD163 and CD206 (mannose receptor).

The M (Hb) macrophage phenotype is induced in vitro by HH complexes, and is characterized by the production both anti-inflammatory (IL-10, IL-1Ra) and pro-inflammatory (VEGF, IL-1β) factors. As compared to foam cells, M (Hb) macrophage lacks lipid retention which distinguishes M (Hb) macrophage from foamy macrophages [71, 77]. Owing to increased activity of the transcription factor oxysterols receptor LXR-α (also known as liver X receptor-α), induction of cholesterol efflux, and downregulation of scavenger receptors involved in lipid uptake these macrophages are protected against lipid accumulation [71]. As eluded to above, free iron is exported through upregulation of the iron exporter FPN in M (Hb) macrophage, which leads to a decrease of intracellular free iron in M (Hb) macrophage [71, 78]. M (Hb) macrophages also produce less reactive oxygen species (ROS) than foam cells [71]. In response to HH complex ligation via CD163, these macrophages produce IL‑10 by a mechanism involving the phosphoinositide 3-kinase (PI3K)–Akt pathway [79, 80]. On the basis of their lower ROS, elaboration of anti-inflammatory cytokines and non-foam cell appearance, M (Hb) macrophage has thought to be atheroprotective based on these findings [77, 81]. However, the association of M (Hb) macrophage with areas of intraplaque angiogenesis and permeability raises the important question of whether M (Hb) macrophage causes these processes or is merely associated with them.

Vascular endothelial growth factor (VEGF) is a critical signal protein produced by a variety of cells that stimulates the formation of blood vessels by binding to tyrosine kinase receptors (VEGFR) in the cell surface [82]. The production of VEGF is stimulated by hypoxia-inducible factors (HIFs). A relationship between iron and HIF1α can be found in its interactions with the prolyl hydroxylase proteins (PHDs). In the milieu with normoxia or hyperoxia, HIF1α becomes hydroxylated by PHDs on proline residues 402 and 564, within its oxygen-dependent domain. This allows it to be recognized by the von Hippel–Lindau tumor suppressor protein (pVHL), which targets it for ubiquitin-mediated degradation [83]. Because it is an essential cofactor for the activity of PHDs, iron also indirectly controls the activation of HIFIα [84]. The hydroxylation of proline residues on HIF1α is controlled by 3 HIF prolyl hydroxylases (PHD1, PHD2, PHD3) [85]. PHD2 plays a primary role in cultured cells [85]. The addition of iron in vitro promotes PHD2 hydroxylase activity whereas its chelation is associated with PHD inhibition. Thus, the angiogenic potential of a cell can be affected by iron metabolism. We recently showed that lowering of intracellular iron within M (Hb) macrophages inhibits PHD2 function which increases HIF1 and VEGF levels. In vitro work suggested M (Hb) supernatants increased endothelial permeability and endothelial inflammation [i.e., upregulation of vascular cell adhesion molecule (VCAM)] which was also shown by comparing plaque progression in the brachiocephalic artery of CD163+/+ and CD163−/− mice both on ApoE−/− background (Fig. 3) [82]. Blockade of VEGF both in vitro and in vivo inhibited these effects suggesting VEGF as the main stimulus. In human plaque the presence of CD163+ macrophages was correlated with vascular permeability (as measured by Evans Blue Dye staining), angiogenesis and inflammation. Finally, in 1-year-old CD163−/− ApoE−/− brachiocephalic artery plaque progression was inhibited as compared to CD163+/+ ApoE−/− mice, suggesting CD163 is a proatherogenic molecule.

Fig. 3
figure 3

Summary of the role of M (Hb) macrophages in plaque angiogenesis, permeability, vascular inflammation, and plaque progression. In areas of IPH, HH complex ingestion by macrophages induces angiogenesis via activation of HIF1α, which is a consequence of intracellular Fe2+ deprivation and PHD2 inhibition. VEGF-A, which is secreted by macrophages via HIF1α activation, promotes angiogenesis, endothelial expression of VCAM, inflammatory cell recruitment, and vascular permeability via VEGF-A/VEGFR2 signaling. This may cause further IPH, RBC lysis, and more Hb ingestion by CD163+ macrophages. This vicious cycle causes plaque progression, which eventually leads to plaque rupture (Ref. [82])

Full size image

These results challenge the concept that M (Hb) is atheroprotective and that hypoxia resulting from intimal thickening is the major cause of plaque angiogenesis. The relationship between hypoxia in human carotid artery and the bioreductive agent pimonidazole was reported by Sluimer et al. [86]. Hypoxia was strongly linked to regions of advanced atheroma with the macrophage marker CD68, angiogenesis, and thrombus, though pathological intimal thickening lesions are not related to hypoxia. At the sites of inflammation, HIF1α immunoreactivity and mRNA were detected even in areas within a close distance (20–30 μm) from the vessel lumen though the diffusion limit of oxygen is 100–250 μm. While hypoxia may be an additional and important factor, our data clearly demonstrate that Hb intake by human macrophages under conditions of normoxia drives their differentiation into proangiogenic cells.

These data suggest that the proangiogenic changes that occur within M (Hb) cells are almost certainly proatherogenic within the context of IPH. Small amounts of intimal angiogenesis and bleeding could be exacerbated by the inflammatory response which sets in place a vicious cycle by which bleeding begets more bleeding. Clinical studies using carotid MRI to detect IPH also confirms this event has long-standing deleterious effects on plaque growth that remain detectable for years after the initial bleeding event [87].

Role of macrophages in plaque stability

As the lesion becomes advanced, the number of macrophages increases. Symptomatic plaque has more macrophages than asymptomatic plaques [88,89,90]. Macrophages in shoulder regions of fibrous cap mainly express M1 polarization markers and these are recognized as unstable areas prone to cap rupture [91]. Macrophages found in fibrous plaque express both M1 and M2 polarized markers [88]. Ruptured human plaques express transcriptional markers of both M1 and M2 cells without preferential polarization. Ruptured versus stable plaques showed upregulation of both M1 markers such as CD68, HLA-DP/Q/R and MARCO and M2 markers DC-SIGN and PPARγ. Markers of IPH such as CD163 and HO-1 were very upregulated in ruptured plaques. Immunostaining revealed both types of markers accumulate as plaques increased as lesion severity increased. Lesional foam cells were characterized by some markers of both M1 (HLA-DP/Q/R and iNOS as well as dectin-1).

According to some experimental data, macrophage phenotype can switch to differential subtypes in response to the microenvironment. These might explain the reason why macrophage phenotypes expressing different types of markers exist in same regions or the fact that foam cells demonstrate mixed of surface markers. Mouse studies have reported phenotype switching of macrophages [92]. Arginase II (M1 markers) has been found next to arginase I (M2 markers) in macrophages within advance atherosclerotic plaques. Moreover, immunohistological staining showed equally distributed arginase I and II within advanced plaques [92]. Switching phenotypes of macrophages can be influenced by changes of cytokines in microenvironment with plaque progression and severity.

Atherosclerosis development in coronary arteries can be also be affected by macrophages in epicardial adipose tissue. There are more macrophage infiltration and cytokine expression in the epicardial adipose tissue of patients with coronary artery disease than those without coronary artery disease [93, 94]. The M1:M2 macrophage ratio in epicardial adipose tissue is positively correlated with the severity of coronary artery disease, indicating that epicardial adipose tissue as well as the plaque are related to macrophage polarization in atherosclerotic lesion of humans [94].

M1 macrophages are dominant within the carotid plaque of symptomatic patients during acute ischemic attack, whereas asymptomatic patients are likely to display M2 phenotype in the plaque [89]. CD11c + M1 macrophages were found within the advanced plaques related to acute myocardial infarction. On the other hand, there were no differences in the mannose receptor + M2 macrophages between acute myocardial infarction lesions and stable angina lesions [95]. Moreover, the different types of arteries are also associated with proportion of M1 and M2 phenotypes [90]. M1 macrophages are more frequently observed in carotid arteries than femoral arteries, whereas M2 markers were detected more common in femoral than carotid arteries [90]. The results of correlation studies indicate that M1 macrophages predominate in plaques with an unstable phenotype, suggesting that plaque instability might be the consequence of an imbalance between M1 and M2 phenotypes. However, whether the abundance of M1 macrophages in unstable plaques is the cause or consequence of plaque rupture remains unclear and further work is needed to understand how the ratio of macrophages phenotypes affects plaque progression.

Atherosclerosis regression

Historically, the progression of atherosclerosis in both human and animal studies has been the focus of most research. It was a long-held belief that atherosclerosis was essentially an irreversible process, though plaque regression remained poorly understood. The most convincing evidence supporting regression in advanced lesions has come from animal studies. These experiments date back to the 1920s when Anichkov showed that switching cholesterol-fed rabbits to a low-fat chow over 2–3 years resulted in arterial lesions becoming more fibrous with reduced lipid content [96]. Subsequent studies in monkeys showed shrinkage and remodeling during longer follow-up as their diet was low-fat diet [97]. To further understand the mechanisms of plaque regression, murine transplantation models have been used. In these experiments, a segment of plaque containing aorta from a hyperlipidemic proatherogenic (e.g., ApoE−/−) is transplanted into a wild-type recipient. Using this approach the degree of plaque complexity can also be examined. Transplanting early or advanced lesions into wild-type mice reduces the number of foam cells and increases the number of smooth muscle cells, particularly in the fibrous cap [98, 99]. Recent studies showed that continued recruitment of Ly6Chi inflammatory monocytes and their STAT6-dependent polarization to the M2 state are required for resolution of atherosclerotic inflammation and plaque regression, and single-cell RNA-seq analyses showed less complex macrophage populations in the regression than progression [38, 100].

Common to these mouse models of atherosclerosis regression is a decline in the number of macrophages and some changes in their phenotypic characterization with enrichment of M2 characteristics, suggesting these cells may play an active role in this process. As more data accumulated for the existence of multiple macrophage phenotypes in atherosclerosis, the emigration of macrophages and the presence of tissue-remodeling M2 macrophages in human and animal plaques were reported. They raise the hope that atherosclerosis regression can be achieved at the clinical level. In progressing and regressing plaques of an aortic transplantation mouse model, transcriptomic profiling of macrophages isolated by laser capture microdissection showed > 700 differentially regulated genes [31]. Downregulated genes included adhesion molecules, such as members of the cadherin family [31] with the upregulation of cellular motility factors. Transcriptome analyses from other models of regression will be needed to determine how conserved these changes are.

Because it remains impossible to detect macrophages phenotypes in humans, the association of M2 macrophages with plaque regression in living patients remains uncertain. Clinical trials in which patients with acute coronary syndromes as well as stable angina were treated with high-dose statins and evaluated by intravascular ultrasound (IVUS) have shown changes in plaque composition [101, 102]. However, meta-analysis from eight pooled clinical trials evaluating the effects of intensive statin therapy on plaque volume by IVUS showed that high-intensity statin therapy significantly increases the volume of calcification within plaque without a decrease of atheroma volume [103]. Additional clinical studies also support these results [104, 105]. Taken together, statin therapy may provide plaque stabilization with volume reduction of atheroma with some incremental increase in calcification volume. Yet it remains unknown how such treatments affect plaque macrophage phenotypes in humans.

Conclusions

During the development and progression of atherosclerotic lesions, macrophages respond to various environmental signals such as lipid and their derivatives, pro- and anti-inflammatory cytokines and heme from senescent erythrocytes. They can regulate distinct phenotypes of macrophage in atherosclerotic lesions. The spectrum of macrophage phenotypes probably changes over the lifetime of a plaque since these signals may vary as plaques progress or regress depending upon the cellular events taking place. Therefore, macrophages adapt their phenotype both over time and in response to what is taking place around them. Although it is well recognized that foam cell type macrophages play a critical role in atherosclerosis progression, the role of function of alternative macrophage phenotypes is still being explored. In fact, the recent single-cell-RNA-seq studies revealed that the foamy macrophages express lesser inflammatory cytokines discrepant with the commonly held perception that foamy macrophages are related to the inflammatory burden. While IPH is a well-recognized event in the life of advanced plaques, the macrophage inflammatory response to such an event may also be an important predictor of plaque evolution. The role of M2 macrophages as associated with plaque regression in humans is still not fully understood. The concept of therapeutic targeting of macrophages within atherosclerotic lesions is attractive as a method to prevent the progression or even hasten regression of plaques. However, modulation of macrophage phenotypes as a therapeutic approach remains an attractive concept but is not yet feasible yet since several questions about the role of alternative macrophages in plaque evolution remain unanswered.