Macrophages as tissue scavengers, builders, and sensors

Abstract

The cellular biology of macrophages underpins the multitude of roles that these cells undertake under homeostatic and inflammatory conditions. Macrophages populate all tissues where they contribute to organ physiology while acting as sensors of health and triggering inflammation in response to organ dysfunction, trauma, and infection. Sharing key characteristics such as a highly developed endocytic compartment, secretion of growth factors and cytokines, motility and antigen presentation, macrophages undergo specific adaptations in each niche guided by environmental clues that result in diverse phenotypes that support tissue-specific roles such as iron recycling, synaptic pruning, bone reabsorption, and processing of lung surfactant. This review will provide an overview of macrophage biology and heterogeneity that underpin their contribution to homeostasis and inflammation to illustrate their importance as therapeutic targets in a wide range of inflammatory diseases.

THE FAMILY OF MONONUCLEAR MYELOID CELLS

Macrophages are members of the mononuclear myeloid cell family that include monocytes and dendritic cells. Macrophages and dendritic cells (DCs) are in all tissues, acting as major sensors of their environment with macrophages specialized in clearance, maintenance of tissue homeostasis (1), and instructing inflammation (2) and dendritic cells specialized in priming naïve T cells in secondary lymphoid organs. Monocytes circulate in blood and can differentiate into macrophages and dendritic cells (mo-macrophages and mo-DCs) upon migrating into tissues (3).

MACROPHAGES AS PROFESSIONAL PHAGOCYTIC CELLS

The Phagocytic Process

The most prevalent characteristic of macrophages and the other members of the mononuclear myeloid cell family is their ability to internalize particulate material (≥0.5 µm) through phagocytosis. The range of targets for macrophages’ incredible appetite is extensive and encompasses, among others, microbes of a wide range of sizes and from different groups, including bacteria, fungi, protozoa, and viruses, apoptotic cells and particulate pollutants. Cargo recognition can be direct through nonopsonic receptors or through opsonic receptors that include the receptors for the Fc portion of immunoglobulins (FcRs) and for complement fragments such as CR3 (CD11b/CD18) and CR4 (CD11c/CD18) that recognize iC3b (4, 5). The vacuoles containing the internalized particles are called phagosomes and undergo a maturation process characterized by sequential fusion events with early endosomes, late endosomes, and lysosomes that change the composition of the phagosome membrane and lumen (6). Hence maturing phagosomes go through different stages, termed early phagosomes, late phagosomes, and phagolysosomes, acquiring hydrolases and peptides and proteins with direct bactericidal activity, such as defensins and cathelicidins, or bacteriostatic activity through nutrient deprivation such as lactoferrin, that mediate the degradative and antimicrobial activities of the phagosomes. Increasing levels of the proton pump V-ATPase causes acidification with pH reducing from 6.1–6.5 in early endosomes to 5.5–6.0 in late endosomes and 4.5 in phagolysosomes (6). Production of reactive oxygen and nitrogen species through the action of NOX2 NADPH oxidase and the inducible nitric oxide synthase NOS2, respectively, further aid damaging of macromolecules in the phagosomal cargo by interacting with thiols, metal centers, tyrosine residues, nucleic acids, and lipids (6). The uptake of soluble material by macrophages, endocytosis, follows a parallel process and internalized solutes are packaged into endosomes that undergo maturation and final fusion with lysosomes (5).

In macrophages, phagocytosis and endocytosis are underpinned by a highly developed endosomal compartment alongside microtubule-mediated vesicle transport and actin nucleation. Polymerized actin accumulates in the phagocytic cup (5), with advancing pseudopods supported by actin polymerization, whereas depolymerization takes place at the base of the phagocytic cup (7). In contrast, neutrophils, another professional phagocytic cell, lack a developed endosomal system, and phagosome maturation occurs through fusion of the phagosomes with preformed primary, secondary, and tertiary granules (7). Vesicle recycling is essential to maintain phagosome size and removal of specific contents and takes place through “fission complexes” that induce the formation of budding vesicles (5). Plasma membrane surface increases during phagocytosis through the process of “focal exocytosis,” with multiple compartments contributing, including recycling endosomes, late endosomes, and lysosomes (8), showcasing the ability of macrophages for rapid membrane remodeling.

Disposal of Phagocytic Cargo

Products of phagosomal degradation are further processed through different disposal mechanisms. For instance, amino acids resulting from protein degradation are translocated across the membrane by solute carriers. Lipids are extracted from the membrane of the cargo through the action of lipid transfer proteins, such as saposins, to enable digestion by lipases. Nucleic acids are digested by DNAse II, and nucleotides will need to be dephosphorylated to generate nucleosides that exit lysosomes through the equilibrate nucleoside transporter 3 (ENT-3) (7).

Immunological Consequences of Phagocytosis

The nature of the phagocytic cargo will impact the fate of phagosomes and macrophage responses. For instance, macrophages will readily internalize apoptotic cells (efferocytosis) through recognition of molecular cues displayed at the surface of apoptotic cells through specialized receptors, thereby becoming immunosuppressive producing IL-10 and transforming growth factor (TGF)-β (9). Phosphatidylserine (PtdSer) is a key “eat me signal” for apoptotic cells, and multiple receptors for PtdSer have been identified that bind directly such as brain-specific angiogenesis inhibitor BAI-1 and T cell immunoglobulin and mucin domain-containing proteins (TIM-1 and TIM-4). Other receptors include promiscuous scavengers that bind negatively charged lipids, apolipoproteins, and bacteria, among others (10). PtdSer can also be recognized through bridging soluble molecules such as milk fat globule-EGF factor 8 (Mfge-8) and Gas6 that bind PtdSer and receptors on macrophages. The existence of multiple PtdSer receptors might reflect cell specificity and differential involvement during phagocytic uptake (10).

During the uptake of apoptotic cells, there is recognition of oxidized fatty acids and increased levels of cholesterol, and peroxisome-proliferator-activated receptors (PPAR-δ and PPAR-γ), and liver X receptor (LXR), which sense modified fatty acids and sterols, are activated and contribute to the anti-inflammatory response of macrophages (11). The combination of signals triggered by apoptotic cells and those triggered by microbial compounds changes the immunological outcome. For instance, infection with bacteria expressing apoptosis-inducing toxins that tend to cause substantial tissue damage leads to development of Th17 cells that could promote tissue repair through IL-22 (11).

In addition to timely removal of apoptotic cells and senescent erythrocytes, there are remarkable examples of macrophage phagocytosis supporting tissue homeostasis. For instance, cardiac macrophages sustain cardiomyocytes by eliminating excreted damaged mitochondria (12). Bone marrow macrophages support maturation of erythroblasts by eliminating nuclei and mitochondria extruded by these cells. The recognition of extruded nuclei (termed pyrenocytes) involves PtdSer and Mfge-E8 (13, 14).

Effect of Macrophage Activation on Phagosome Maturation

Finally, phagosome maturation will depend on macrophage activation state. As example, one study investigated human macrophages generated from monocytes incubated with GM-CSF or macrophage-colony stimulatory factor (M-CSF) for 5 days and additional 2-day treatment with either lipopolysaccharide (LPS) together with IFN-γ or IL-4 (considered by the authors M1 and M2 macrophages, respectively). See discussion on macrophage activation in macrophage heterogeneity (15). GM-CSF-IFN-γ-LPS macrophages had higher NOX2 activity, leading to high superoxide generation with protons being consumed at high rate. This, together with reduced recruitment of the V-ATPase to phagosomes prevented acidification. In contrast, in the M2 macrophages, phagosomes were more acidic as proton pumping was enough to compensate for dismutation (7, 15). In the mouse system using bone marrow-derived macrophages, the rate of phagosome-lysosome fusion was initially delayed in the presence of IFN-γ alone or in combination with LPS, but the process was maintained for longer, leading to phagolysosomes with a higher concentration of lysosomal constituents. This delay in phagosome maturation was accompanied by delayed acquisition of V-ATPase hence reducing the rate of phagosome acidification. There was also reduction in phagosomal proteolytic activity (16). In contrast, macrophages generated in the presence of IL-4 had increased proteolytic machinery (cathepsin S and L) and decreased phagosomal NOX2, leading to a reducing microenvironment that promotes activities of local cysteine cathepsins and ability to reduce disulfide bonds (17). Pathogens exploit the versatility of the phagocytic process to survive within macrophages by modifying the phagosome maturation process or escaping to the cytosol (6).

MACROPHAGES AS SENSOR AND SCAVENGERS

Macrophage Receptors

The endocytic and phagocytic activities of macrophages make them important scavengers and sensors in tissues. Cargo recognition and internalization and ensuing cellular responses are mediated by a variety of receptors that dictate phenotypic changes in the cell by engaging specific signaling pathways. Hence, unlike T and B cells, whose activation is vastly dependent on a single and highly specific receptor per cell, restricted to secondary lymphoid organs, and delayed, macrophages are suited to respond rapidly to environmental changes, including tissue damage, metabolic stress, and the presence of microbial compounds, by expressing numerous receptors in a single cell. These receptors, rather than having high specificity, recognize molecular patterns associated with normal homeostasis (i.e., apoptotic cells), trauma (i.e., extracellular matrix fragments and intracellular material), and infection (i.e., LPS, a major component of gram-negative bacteria, and β-glucan, a component of fungal cell wall) (18, 19).

This paradigm of pattern recognition has been instrumental in defining how macrophages, key components of the innate immune system, trigger fast and, to some extent, tailored responses by integrating signals from multiple, simultaneously, or sequentially engaged pattern recognition receptors (PRRs). In general, molecular patterns that trigger an inflammatory response are associated with tissue damage, i.e., damage-associated molecular patterns (DAMPs) or presence of infection, i.e., microbe-associated molecular patterns (MAMPs), commonly referred to as pathogen-associated molecular patterns (PAMPs). As mentioned earlier, the molecular patterns associated with homeostasis, i.e., apoptotic cell clearance, lead to immune suppression. Other examples of receptors that dampen macrophage activity are CD47 that acts as a “do not eat me” signal by engaging its counter receptor SIRP-1α. The CD47-SIRP-1α interaction is considered as an immune checkpoint in cancer. Sensing of DAMPs alongside MAMPs explains how the immune system discerns between commensals and pathogens. Although both groups share MAMPs, in general, only pathogens express virulence factors that cause tissue damage, i.e., DAMPs (20).

Pattern recognition occurs at the plasma membrane, endosomes, and cytosol, and is mediated by receptor families with diverse domain composition and signaling capabilities being the best studied toll-like receptors (TLRs), C-type lectin receptors (CLRs), scavenger receptors, NBD-containing and leucine-rich repeat (LRR)-containing protein (NLRs), RIG-1-like receptors (RLRs), and cGAS-STING. This review will offer a general overview of their main characteristics.

Toll-Like Receptors

TLRs (21, 22) encompass a family of 10 receptors in humans sharing the overall structure, an N-terminal domain consisting of Leucine-rich repeats (LRRs) that mediate ligand binding, a transmembrane domain, and a C-terminal Toll/interleukin-1 receptor domain (TIR) domain that triggers signal transduction, leading to activation of the major transcription factor NF-κB and, depending on the TLR family member, of the type I IFN pathway. The TIR domain is also present in the cytoplasmic tail of the receptors for the IL-1 superfamily, which includes the alarmins IL-1α and IL-33 and IL-1β and IL-18 (23, 24), hence these cytokines, routinely produced early during inflammation, could be considered as endogenous enhancers of innate immune activation. Receptor dimerization is important for TLR ligand binding and signaling, and TLRs form either homodimers (TLR3, TLR4, TLR5, TLR7, TLR8, TLR9, and TLR10) or heterodimers (TLR1/TLR2 and TLR6/TLR2) (25).

TLRs at the plasma membrane detect lipoproteins (TLR1/TLR2, TLR6/TLR2), LPS [TLR4, together with MD2 and CD14 (26)], and flagellin (TLR5) (27) and trigger mainly NF-κB signaling, leading to the synthesis of inflammatory cytokines, including TNF-α and IL-6 (28). In contrast, endosomal TLRs (TLR7, TLR8, and TLR9) detect nucleic acids (29) and also trigger type 1 IFN production. This agrees with nucleic acids, the most important viral MAMP, inducing synthesis of a key virus-restricting pathway. TLR-4 is unique in triggering a second signaling cascade after being internalized into endosomes following ligand binding, a process controlled by CD14 (25). As for other endosomal TLRs, TLR4 in endosomes drives type 1 IFN responses (21) and has been implicated in viral recognition (30). TLR10 is poorly understood and it is suggested that it promotes anti-inflammatory responses (31).

Lectin Receptors

Differences in carbohydrate composition contribute to sensing of tissue health (32). For instance, disturbance of protein glycosylation leads to SLE-like symptoms in mice (33) and molecules displaying sialic acid, a carbohydrate normally present at the surface of healthy cells, promotes immune suppression through engagement of Siglecs (34). In contrast, mannose, more abundant at the surface of pathogens and in inflammatory molecules that require fast clearance, is a signal for elimination. Two important mannose- and fucose-binding cell surface lectins, such as CD206 (the mannose receptor) and dendritic cell-specific intercellular adhesion molecule-3-grabbing non-integrin (DC-SIGN, CD209), lack signaling motifs and are largely involved in endocytosis and cell adhesion (32, 35), which agrees with their ability to recognize both endogenous compounds and pathogens. Both can modulate cellular signaling when engaged alongside signaling PRRs. The carbohydrates within the fungal cell wall represent major patterns for the detection of fungal infection. Accordingly, multiple lectin receptors have been implicated in fungal recognition in addition to CD206 and DC-SIGN (36, 37). These include the mannose-binding receptor dectin-2 and the β-glucan receptor dectin-1 [C‐type lectin domain family 7 member A (CLEC7A)]. Both receptors signal through the kinase Syk and the adaptor protein caspase recruitment domain 9 (CARD9) complexed with mucosa-associated lymphoid tissue lymphoma translocation protein 1 (MALT1) and B-cell lymphoma 10 (Bcl10), leading to NF-κB activation (36). Although Dectin-2 signals by interacting with the immunoreceptor tyrosine‐based activation motif (ITAM)-containing adaptor FcR common gamma chain (FcRγ), Dectin-1 has an incomplete ITAM (termed hemiTAM) at its cytoplasmic tail. The importance of signaling lectin receptors for protection against fungi is illustrated by the increased susceptibility of CARD9-deficient individuals to fungal infection (38). Some lectin receptors bear immunoreceptor tyrosine-based inhibitory motif (ITIM), leading to recruitment of phosphatases and dampening of inflammation, and these include the Siglecs CD33 and Siglec7 (human) and the C-type lectins myeloid inhibitory C-type lectin-like receptor (MICL, Clec12a) (39). Balance between activating and inhibitory signals will establish the extent of cellular responses as shown for Fc receptors (40). Some C-type lectin-like receptors do not recognize carbohydrates. These include Celc12a, which binds uric acid (39), MINCLE, which, in addition to glycolipids of microbial origin, recognizes the alarmin Sin3-associated protein 130 (SAP130), β-glucosylceramide, cholesterol sulfate, and cholesterol crystals, promoting inflammation in response to tissue damage (41, 42) and dendritic cell natural killer lectin group receptor-1 (DNGR-1, CLEC9A), which recognizes filamentous actin (F-actin) exposed on dead cells (43).

Scavenger Receptors

Scavenger receptors (SR) were first identified as receptors for modified low-density lipoprotein (LDL) and currently there are eight classes (A–H). They were first described in the context of atherosclerosis and hypercholesterolemia. Members of this family are structurally diverse and have broad ligand-binding specificity. They can act as adhesion and endocytic receptors, and some collaborate with TLRs for the induction of inflammation (44). Members of this family include Class A SRs, SR-A1 (CD204) and macrophage receptor with collagenous structure (MARCO, SR-A2), CD36 (45), and CD163 (46).

Cytosolic Pattern Recognition Receptors

NLRs (47) are cytosolic and include the peptidoglycan receptors NOD1 and NOD2 that trigger NF-κB signaling through the kinase RIPK2 (48) and have been also implicated in sensing cellular homeostasis (49). Some members of the NLR family oligomerize after ligand sensing, driving formation of macromolecular structures called inflammasomes, which activate the inflammatory caspase 1 (50). Caspase 1 processes pro-IL-1β and pro-IL-18 and gasdermin-D, leading to secretion of processed, active IL-1β and IL-18 through pore formation by the N-terminus of gasdermin D (50, 51). Pore formation by activated gasdermin D can lead to a form of inflammatory cell death called pyroptosis with the release of DAMPs and proinflammatory cytokines (50). Gasdermin D can be processed by another inflammatory caspase (caspase 11 in mice and caspases 4 and 11 in humans) after sensing intracellular LPS through the formation of a noncanonical inflammasome (50, 52).

Finally, RLRs (53), RIG-1 and MDA5, are cytosolic RNA sensors (54), and stimulator of interferon response cGAMP interactor 1 (STING1) is a sensor of the cyclic dinucleotide 2′3′ cGAMP, synthesized by cGAS in response to cytosolic dsDNA (55, 56). Both systems promote synthesis of type 1 IFN in addition to NF-κB as seen previously for endosomal TLRs, again highlighting the importance of nucleic acid recognition for sensing of viral infection.

As mentioned earlier, macrophages express opsonic receptors for fragments of activated complement C3 (CRs1–4) and the Fc region of IgG (FcRs), and FcRs can be either activating or inhibitory (40). Following on the concept of signal integration, it is likely that nonopsonic PRRs and CRs contribute to early immune activation and, once specific IgGs are present, all these receptors contribute to cellular uptake and modulation of cellular activation. Figure 1 illustrates general characteristics of macrophages.

Figure 1.

Macrophages as sensors of molecular patterns within tissues. Macrophages determine the health status of their surroundings through pattern recognition. Healthy tissues will establish interactions with macrophages that inhibit inflammation and sustain organ function. Under pathological conditions due to infection (presence of MAMPs) alongside cell stress or damage (presence of DAMPs), macrophages will promote inflammation. Pattern recognition occurs through receptors at the plasma membrane, endosomes/phagosomes, cytosol, and nucleus. Signal integration from multiple receptors will lead to tailored responses. Importantly, the threshold (dashed gray line) and nature of the macrophage responses will be influenced, among others, by epigenetic modifications acquired through tissue adaptation of early precursors during development or recruited monocytes after birth and previous exposure that can cause a tolerant state (reduced response) or a trained state (enhanced response). Finally, the dominant immune status during inflammation, established through interactions with activated T cells, antibodies, and granulocytes, further modulate macrophage behavior. DAMPs, damage-associated molecular patterns; MAMPs, microbe-associated molecular patterns.

MACROPHAGE HETEROGENEITY

Historical Perspective

As in many aspects of immunology, the study of macrophage biology greatly benefits from rodent models, where it is possible to sample healthy and diseased animals at different stages of development and after birth. In humans, the vast majority of the studies on the biology of primary macrophages use monocyte-derived and alveolar macrophages. In mice, researchers exploit bone marrow cultures exposed to macrophage-colony stimulatory factor (M-CSF, CSF-1), which, together with IL-34 and acting through the CSF-1R, promotes the development of most tissue macrophages (57). Other common sources of mouse macrophages are the peritoneal and alveolar spaces. The extraction of macrophages from solid organs requires careful enzymatic digestion to minimize cellular damage and degradation of surface markers and can lead to artefacts, see Ref. 58 as example.

Early use of cellular markers in mice provided the first indication of the heterogeneity of macrophages in tissues (59). For instance, the mouse spleen and lymph nodes contain multiple macrophage populations strategically positioned to support clearance and induction of adaptive immunity (60–62). In addition, analysis of selected cellular markers, such as the mannose receptor (CD206), provided an early indication of how macrophages undergo distinct activation pathways (63, 64), undertaking important physiological functions (65). These original studies also endeavored to establish correlations between expression of a particular subset of markers and cellular activity in cultured macrophage to inform on the potential role of different macrophages in situ. The Th1/Th2 differentiation paradigm led to the exploration of macrophage responses to the canonical Th2 cytokines IL-4 and IL-13 and coining the term “alternative/M2 activation,” whereas macrophage activation in response to the Th1 cytokine IFN-γ and LPS became “classical/M1 activation” (66). In fact, macrophages with a wide range of phenotypes are generated in vitro by varying their source, culture conditions, exposure to different combinations of MAMPs and DAMPs, hormones, and cytokines to model prevalent physiological conditions. This malleability is consistent with their important role as health sensors and effectors and aligns with the concept of macrophages existing within a spectrum of activation states (67) and the heterogeneity of macrophages identified through single-cell analysis (see Epigenetics and Tissue Adaptation and Epigenetics and Macrophage Activation Threshold). Key functional readouts in vitro include microbicidal activity (68), production of cytokines (69), metabolic changes (70–72), interaction with the extracellular matrix (73), and effects on T cells (74).

Epigenetics and Tissue Adaptation

Changes in gene expression through activation of transcription factors in response to receptor engagement, including mechanosensing (73, 75, 76), underpin the heterogeneity of macrophage phenotypes. Some of these phenotypes become further entrenched by epigenetic changes associated with the presence of specific metabolites (77) such that macrophages in different anatomical locations have differing epigenomic landscapes sustained, among others, by histone modifications (78–80). Within the tissue environment, macrophages sustain important homeostatic functions underpinned by their versatile cellular properties (see the family of mononuclear myeloid cells, macrophages as professional phagocytic cells, and macrophages as sensor and scavengers), including synthesis of growth factors, cytokines, and metabolites, and clearance of damaged cells (1). These cellular characteristics also endow macrophages with the capacity to sense changes in their surroundings linked to cellular damage and stress and infection and underpin their central role in driving and regulating inflammation (2, 81–83).

Rodent models established that macrophages are present in all organs (84) and either originate from yolk sack precursors or fetal monocytes seeded during development or are derived from monocytes differentiated from hematopoietic stem cells (HSCs) after birth (85, 86). Examples of embryonically seeded macrophages are alveolar macrophages (87) and microglia (88). Embryonically seeded macrophages are maintained through self-renewal rather than being replaced by migrating monocytes, although this is highly dependent on the organ and inflammatory conditions (86). Initial characterization of tissue macrophages involved the study of pooled preparations of all macrophages present within particular organs, but more advanced analysis at the single-cell level has unveiled the existence of multiple macrophage subpopulations (85) and it is now apparent that macrophages with different origins, locations, and functions coexist. In turn, there are also attempts at identifying core macrophage functions shared by all tissue macrophages (89), such as expression of Csf1r and Maf that ensures survival, and capacity for efferocytosis and opsonic and nonopsonic phagocytosis, which were already present in fetal macrophage precursors (pre-MACs) (90). Importantly, as highlighted by Mass et al. (85), researchers should adopt technologies such as RiboTag and single-cell RNA sequencing (scRNA-Seq) alongside imaging flow cytometry and three-dimensional (3-D) microscopy to ensure the identification of true macrophage subpopulations, avoiding issues with contamination and influence of phagocytosed cargo. The knowledge of the origin and function of tissue macrophages is ever expanding (1, 85, 86, 91), and remarkable examples of macrophage contribution to tissue homeostasis are constantly reported. For instance, in addition to the elimination of mitochondria excreted by cardiomyocytes (12) and nuclei excreted by differentiating erythroblasts (14) mentioned earlier, macrophages facilitate electrical conduction in the heart (92) and excite muscle spindles with glutamate to support locomotion (93). There remains challenges such as lack of understanding of how macrophages change throughout life and in response to infection and cancer and during chronic diseases. This is important as local insults can influence macrophage populations in distant organs not subjected to the initial damage (94). The contribution of monocyte-derived macrophages to the pool of tissue macrophages increases with age leading to steady replacement of embryonically seeded macrophages, which has implications for the study of aging and associated comorbidities. See Table 1 for main characteristics of macrophages in several tissues.

Table 1.

Main characteristics of selected tissue macrophages

Organ	Main Characteristics
CNS
Microglia	Located in brain parenchyma. Adult population derived from fetal progenitors and maintained through self-proliferation. Require CSF1R ligands M-CSF (produced by neurons) and IL-34 (produced by astrocytes) for maintenance. Contribute to neural network maintenance, guiding of vessel growth and synaptic pruning among others.
CNS-associated macrophages	Located in CNS interfaces (ventricles, meninges and perivascular space). Most fetally derived. Skull and vertebrae bone marrow could serve as source of mo-macrophage progenitors during inflammation.
Bone marrow
Erythroblastic island macrophages	Required for erythrocyte development, including clearance of expelled nuclei. Regulate osteoblast function
Osteomacs
Osteoclasts	Multinucleated cells. Initially all nuclei embryonically derived, but integration of HSC-derived monocyte nuclei occurs over time. Responsible for bone resorption. Deficiency leads to osteopetrosis. Indirectly contributing to maintenance of HSC niche. Proposed to enhance HSC mobilization, retention and differentiation.
Other	Populations implicated in removing apoptotic immune cells, phagocytosis of opsonized cells.
Liver
Kupffer cells	Embryonically-derived. Line liver sinusoids and exposed to blood flow from portal vein. Mostly involved in elimination of cell debris, immune defense and induction of tolerance. Phagocytose senescent red blood cells. Involved in iron and cholesterol metabolism.
Liver capsular macrophages	Comprise 10% of liver macrophages, partly monocyte-derived.
Other	Central vein macrophages and lipid-associated macrophages also described.
Lung
Alveolar macrophages	Located in alveoli. Most abundant macrophage population in the lung. Populated during embryogenesis, high capacity for self-renewal. Replaced by monocyte-derived cells during aging and inflammation. Important contribution to immune defense and homeostasis. Development-dependent on GM-CSF produced by type II airway epithelial cells. GM-CSF signaling deficiency causes pulmonary alveolar proteinosis.
Interstitial macrophages	2–3 populations depending on the expression of surface markers. Nerve-associated and blood vessel-associated populations. Producers of IL-10.
Gut
Lamina propria-associated macrophages	Major contributors to homeostasis of intestinal barrier. Most fetally derived then replaced by mo-macrophages recruited though CCR2. Require microbiota for full development. Existence of long-lived subpopulation that expresses TIM4. Producers of IL-10 and TGF-β. Support epithelial stem cell proliferation in intestinal crypts through production of Wnt ligands. Important for induction of regulatory T cells specific from microbiota.
Other macrophage populations located in submucosa and muscularis externa.	Muscularis macrophages implicated in the maintenance of intestinal movement.
Skin and oral mucosa
Langerhans cells	Located in epidermis. Essential for maintenance of skin barrier. Present in oral mucosa, cornea, and mucosa of reproductive female reproductive tract.
Dermal macrophages
Vessel-associated macrophages	Contribute to maintenance of vessel integrity and with anti-fibrotic activity.
Sensory-nerve-associated macrophages	Contribute to nerve regeneration after injury.
Spleen
Red pulp macrophages	Phagocytose senescent erythrocytes, degrade hemoglobulin, store iron as ferritin and release iron back to the circulation where it binds to transferrin. M-CSF-dependent and M-CSF-independent populations.
MZM	Both populations exposed to circulation in the marginal sinus. MZM located in the outer marginal zone, next to the red pulp, and MMM in the inner marginal zone, next to the white pulp. Take up viral and bacterial pathogens. MMM are implicated in B cell biology and are permissive to bacteria replication.
MMMs
Tingible body macrophages	Located in close vicinity to follicular dendritic cells. Mediates uptake of apoptotic B cells during the germinal center reaction.
Lymph nodes
Subcapsular sinus macrophages	Localized below the subcapsular sinus, adjacent to B cell follicles. Implicated in antigen delivery to follicular dendritic cells and promote B cell activation
Medullary macrophages	Involved in clearance, survival of short-lived plasma cells, and exit of cells and antibodies through the efferent lymphatics. Sinus-associated and cord associated have been described.
Thymus
Thymic macrophages	Clearance of apoptotic thymocytes produced during T cell selection.

Note that most of the data derive from mouse studies using limited number of strains and under specific pathogen-free conditions (85, 95, 96). CNS, central nervous system; HSC, hematopoietic stem cell; M-CSF, macrophage-colony stimulatory factor; MMM, marginal zone metallophilic macrophages; MZM, marginal zone macrophage.

Epigenetics and Macrophage Activation Threshold

Epigenetic changes also underpin altered responsiveness of macrophages and monocytes to stimulation. For instance, extended exposure to low levels of LPS leads to a transient state of tolerance such that macrophages become LPS unresponsive. These tolerized macrophages produce less proinflammatory cytokines when restimulated and express higher levels of negative regulators of PRRs signaling with tolerizeable and nontolerizeable genes identified (97). Tolerance also occurs in response to other TLR agonists, and preexposure to one TLR-ligand will reduce inflammatory responses to another, a process termed cross-tolerance (13, 98, 99).

On the other hand, macrophages can also adopt states of enhanced responsiveness (“training”) in the absence of T and B cells maintained through epigenetic modifications. The term “trained immunity” was coined to account for this novel concept that negates the previous assumption that innate immunity lacked memory (100). Metabolic changes underpin the epigenetic changes associated to training (101). Trained innate memory has been described in natural killer (NK) cells and ILCs in addition to stromal and epidermal stem cells (100). Quintin et al. (102) demonstrated that monocytes recovered from C. albicans-infected mice mediate protection against C. albicans infection when transferred to naïve animals even in the absence of T and B cells. Unspecific protection to infection has been observed in response to live vaccines such as Bacille Calmette-Guérin (BCG), measles, and polio vaccines and has been corroborated in controlled vaccination studies (100). The concept of training innate immunity has now been extended to other infectious stimuli such as malaria (103). Importantly, reprogramming of hematopoietic stem cells, demonstrated in the case of BCG vaccination and β-glucan, confers a mechanism for long-lasting central innate memory (104, 105) that could aggravate chronic inflammatory diseases, in some cases through the action of trained neutrophils (106, 107). Hence, we are facing a scenario where the behavior of myeloid cells from individuals will not be determined just by their genetic makeup and tissue environment but also by past exposure to infection and environmental changes (108), trauma (109, 110), and diet (111) affecting local macrophages and monocytes derived from trained HSC (see Fig. 2 for an overview of factors that impact macrophage heterogeneity).

Figure 2.

Understanding macrophage heterogeneity to guide macrophage-targeted therapeutics. Tissue macrophages originate from early embryonic precursors during development and maintained through replication (light green) or are monocyte-derived (light brown) and adopt specific gene expression profiles guided by the tissue environment. The ratio of both populations varies among organs and during inflammation. The study of human monocytes offers an opportunity for understanding macrophage heterogeneity in humans. Monocytes are themselves highly heterogeneous within and among individuals to a large extent because of environmental factors including microbial exposure, and undergo training at the level of HSCs and can derive either from GMP or MDP precursors. Harnessing the potential of monocyte training offers an opportunity to modulate inflammation but requires extensive analysis to identify suitable subjects to minimise the potential for maladaptive training. GMP, granulocyte-monocyte precursor; HSC, hematopoietic stem cell; MDP, monocyte-dendritic cell precursor.

LOOKING AHEAD: TRANSLATING MACROPHAGE KNOWLEDGE INTO IMPROVED HUMAN HEALTH

We are becoming fully aware of the central role macrophages play in tissue physiology and immunity as sensor and effector cells, as well as the intricacies of macrophage biology and the full extent of their heterogeneity and adaptability. It is daunting to predict how to accelerate the translation of this expanding knowledge to improving human health as adopting any intervention will require extensive knowledge of the prevalent immune status to ensure it is safe and timely.

To study human immunity, numerous studies have used a variety of immune readouts, including serum and plasma biomarkers, numbers, and characteristics of circulating immune cells, as well as their response to in vitro stimulation, to identify immune characteristics that correlate with clinical parameters and to establish a “base line immunity” to stratify patients and healthy volunteers. In healthy individuals, environmental factors rather than genetics dominate immunity (112), including infection history, such as cytomegalovirus infection, and “retraining innate immunity” offers a potential approach to reset human immunity. Although not universally applicable, inflammatory conditions with active recruitment of monocytes might benefit from monocyte preconditioning through HSC training. As mentioned earlier, there are examples of vaccination-induced training that could be exploited for this purpose, but we need to consider that trained immunity can be maladaptive (107). For instance, a training protocol that ameliorates the inflammatory potential of monocytes, without affecting recruitment, could have positive effects in inflammatory tissues such as arthritic joints. Circulating monocytes could act as guiding tools to identify suitable individuals and establish the effectiveness and potential consequences of the retraining process. Toward this aim, detailed knowledge of monocyte biology and heterogeneity is essential.

Considered initially homogeneous, two monocyte subpopulations were identified in humans based on the expression of the LPS coreceptor CD14 and FcR-gamma (CD16), classical (CD14 ++ CD16−) and nonclassical (CD14^loCD16⁺), which correspond to mouse Ly6C^HiCX3CR1^intCCR2+CD62L+CD43^Low and Ly6C^LowCX3CR1^HiCCR2^LowCD62L-CD43+, respectively, with classical monocytes yielding nonclassical monocytes through an intermediate state (113). Extensive phenotypic analysis using multiplex flow cytometry and scRNA-Seq has expanded the number of monocyte subsets, and healthy volunteers have been grouped based on their monocyte subsets (114), and specific monocyte phenotypes linked to chronic inflammatory diseases (115) with a multitude of environmental factors found to influence monocyte characteristics (116). Adaptability is a core characteristic of mononuclear myeloid cells, and these subpopulations could be considered as monocytes undergoing different maturation processes in response to their surroundings. Against this view, some authors have identified functionally different monocyte subsets with limited plasticity predetermined during differentiation. Yáñez et al. (117) demonstrated that mouse monocytes can originate from granulocyte-monocyte precursors (GMPs) or monocyte-dendritic cell precursors (MDPs) in bone marrow and that GMP-derived and MDP-derived monocytes are functionally different with GMP-derived being “neutrophil-like.” Both subsets are preferentially generated during emergency myelopoiesis in response to different microbial stimulation; LPS increased the yield of GMP-derived monocytes and unmethylated CpG DNA increased the yield of MDP-derived monocytes (117). Trzebanski et al. (118, 119) identified similar subsets and demonstrated that GMP-derived and MDP-derived monocytes differed functionally and differentiated into different interstitial tissue macrophages. In a similar line, Rhee et al. (120) established four subsets of mouse monocytes that differ in their ability to respond to gram-negative, gram-positive bacteria (phagocytosis and killing) and produce cytokines in response to LPS that derive from epigenetically programmed myeloid precursors. Importantly, all the subpopulations were able to generate classical and nonclassical monocytes, indicating that this original differentiation pathway occurs independently of the developmental programming of the monocytes. Preestablished monocyte subsets align with findings by Bryson et al. (121) using in vitro-generated monocyte-derived macrophages (Mo-macrophages) cultured in the presence of M-CSF. The authors identified a subset of Mo-macrophages within the cultures able to kill Mycobacteria tuberculosis (Mtb) as those undergoing a GM-CSF-driven signaling, and GM-CSF treatment increased Mtb killing (121). Further support for functional monocyte heterogeneity is illustrated by recent findings, showing that the ability of secreting IL-1β associated with cell death in response to LPS through NLRP3 is restricted to a subset of human monocytes (5%–10%) characterized as CD14^lo CD33^lo cells expressing TLR4, HLA-DR, CD18, and CCR2. CD33 is a negative regulator of TLR4, indicating that TLR4 signaling could be enhanced in the IL-1β-producing monocytes (122).

Finally, it is important not to assume function in vivo based on the expression of particular cellular markers. For instance, tumor-associated macrophages (TAMs) expressing CD206 are generally considered immunosuppressive, but their early deletion in mouse models leads to loss of antitumor immunity, which agrees with CD206+ myeloid cell signatures associating with antitumor immunity in human cancers (123). These findings further illustrate the plasticity of macrophages and are in line with the challenges encountered when targeting TAM (124, 125).

CONCLUSIONS

By assessing current knowledge and state of the art in macrophage biology, this review highlights the immense progress achieved while acknowledging the challenges of understanding and targeting an ever-changing target (i.e., macrophages with vast adaptability and heterogeneity potential). The power of system biology and bioinformatics will guide the identification of best strategies to exploit the potential of manipulating macrophages’ potential to improve human health.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author.

AUTHOR CONTRIBUTIONS

L.M.-P.: Conceptualization, Writing – review & editing.