Key questions and gaps in understanding adipose tissue macrophages and early-life metabolic programming

Abstract

The global obesity epidemic, with its associated comorbidities and increased risk of early mortality, underscores the urgent need for enhancing our understanding of the origins of this complex disease. It is increasingly clear that metabolism is programmed early in life and that metabolic programming can have life-long health consequences. As a critical metabolic organ sensitive to early-life stimuli, proper development of adipose tissue (AT) is crucial for life-long energy homeostasis. Early-life nutrients, especially fatty acids (FAs), significantly influence the programming of AT and shape its function and metabolism. Of growing interest are the dynamic responses during pre- and postnatal development to proinflammatory omega-6 (n6) and anti-inflammatory omega-3 (n3) FA exposures in AT. In the US maternal diet, the ratio of “pro-inflammatory” n6- to “anti-inflammatory” n3-FAs has grown dramatically due to the greater prevalence of n6-FAs. Notably, AT macrophages (ATMs) form a significant population within adipose stromal cells, playing not only an instrumental role in AT formation and maintenance but also acting as key mediators of cell-to-cell lipid and cytokine signaling. Despite rapid advances in ATM and immunometabolism fields, research has focused on responses to obesogenic diets and during adulthood. Consequently, there is a significant gap in identifying the mechanisms contributing metabolic health, especially regarding lipid exposures during the establishment of ATM physiology. Our review highlights the current understanding of ATM diversity, their critical role in AT, their potential role in early-life metabolic programming, and the broader implications for metabolism and health.

INTRODUCTION

The prevalence of childhood and adolescent obesity in the United States has surged nearly 10-fold over the past four decades (1). Presently, nearly 40% of American children are either overweight or obese, contributing to an enormous healthcare expenditure of approximately $14 billion annually (1–3). Alarmingly, one in five children is obese before the age of five, and projections indicate that over 55% of today’s infants will become obese by the age of 35 (1, 4–6). The onset of obesity is not only occurring at younger ages but is also accompanied by a host of adult-associated comorbidities, including type 2 diabetes, nonalcoholic fatty liver disease, cardiometabolic issues, psychosocial consequences, and even premature death (7–9). Childhood obesity strongly predicts obesity in adulthood, often leading to these severe health complications. The sharp rise in childhood obesity underscores an urgent need for innovative research targeting early-life factors that contribute to this epidemic. Given the link between dysfunctional adipose tissue (AT) and numerous noncommunicable diseases, targeting AT to increase energy expenditure is a critical therapeutic strategy to combat obesity.

AT plays an important role in regulating many physiological processes for whole body homeostasis, including insulin sensitivity, hormone production, balance of energy intake and expenditure, thermogenesis, and immune function (10). There are three distinct types of AT depots found throughout the body, each serving unique functions, reviewed extensively by Sakers et al. Briefly, white AT (WAT) is the most abundant and functions in hormone secretion (adipokines) and nutrient storage within its stable, unilocular triglyceride droplets. The major WAT depots include visceral (vWAT) within the peritoneal cavity and subcutaneous (sWAT) beneath the skin (11). In humans, sWAT comprises upward of 80% of total fat mass (12). Beige AT (BeAT) is found in the same depots and shares the same stem-like progenitor cells that give rise to white adipocytes, but they retain the nutrient-burning properties of the ontologically- and anatomically distinct brown AT (BAT) depots. In human infants, the major BAT depot is interscapular and is particularly important for maintaining body temperature during development (13), whereas in adults, BAT is sparse, located in neck and supraclavicular regions, and can be stimulated by cold exposure and small molecule activators (14–16). Importantly, both BeAT and BAT contain abundant mitochondria, multilocular triglyceride storage, and function in thermogenesis through the expression of uncoupling protein 1 (UCP1) (10, 17). Notably, BeAT and BAT can serve as a “metabolic sink” disposing of excess nutrients, able to boost whole body energy expenditure by 40%–80% in humans and more than 100% in mice (17, 18).

Dr. Barker’s (19) fetal programming hypothesis, commonly known as the developmental origins of health and disease (DOHaD) (20), has provided an important framework for our understanding of the early-life molecular, epigenetic, and physiological mechanisms increasing the risk for noncommunicable diseases such as obesity, cardiovascular disease, fatty liver disease, and diabetes (21, 22). Evidence for multigenerational effects from paternal and maternal diets on disease outcomes, even before conception, emphasizes the importance of the DOHaD hypothesis (23–25). Normal fetal development is dependent upon abundant polyunsaturated fatty acids (PUFAs), which are supplied via placental transfer throughout gestation and via breastfeeding in the postnatal window (26). During this early-life period, PUFAs are vital for building cell membranes, producing hormones, and generating signaling lipids, all of which are critical to organogenesis during development. Maternal fatty acid (FA) intake is the main determinant of fetal lipid exposure. For example, excess infant adiposity is more common in offspring from mothers with obesity (25), and has been attributed, in part, to increased expression of placental FA transporters; however, the main driver of fetal absorption of FAs is the concentration gradient between mother and fetus (26).

Recent technological advances, especially single-cell analysis techniques, have significantly facilitated our ability to characterize the diverse populations of cells within AT (5, 6, 10, 27–43). This has revealed that AT is heterogeneous, comprised of adipocytes and their progenitor cells, and various immune cells, such as dendritic cells, natural killer cells, and innate lymphoid cells, each contributing to AT’s multifaceted roles (28, 29, 44–47). It is now firmly established that AT plasticity is, at least in part, influenced by the dynamic responses of AT immune cells to cues from the microenvironment. This interaction facilitates a choreographed process of tissue remodeling, inflammation, and thermogenesis, which varies depending upon the stimuli (10). Macrophages, known for their phagocytic and efferocytic functions during immune responses and tissue remodeling, are increasingly recognized as master regulators of physiological processes including organogenesis, angiogenesis, neuronal function, and adipogenesis (48–51). Studies across many tissues have highlighted the remarkable heterogeneity of macrophages within different niches, revealing both universal cellular characteristics and tissue- and polarization-specific functions. Although various reviews have summarized the tissue-specific functions, ontogeny, and lifespan of macrophages in the brain, liver, and other tissues (52–55), there remains a notable deficit in our understanding of adipose tissue macrophages, particularly regarding early-life programming and function within the developing AT (50, 51). Despite these gaps, our understanding of AT macrophage (ATM) contributions to mitochondrial homeostasis, lipid turnover, efferocytosis, and adipogenesis is well established (48–50). To that end, in vitro and in vivo studies have revealed a close interplay between ATMs and adipocytes, establishing signaling lipids and cytokines as key transducers of intercellular ATM to AT communication. Ligand-receptor analyses suggest that ATMs are central communicators between immune and nonimmune cells in obese WAT (29, 34). Leveraging advanced techniques such as single-cell transcriptomics (29), RNA-sequencing (28), and proteomics (27), recent research offers a comprehensive view of the intricate communication within AT. These studies elucidated an incredibly complex choreography in AT development, homeostasis, and inflammation that is dictated by ATMs.

ATMs are major lipid mediators involved in establishing AT metabolism. The first evidence of tissue-resident macrophages (TRMs) within the AT of obese mice was reported in the 1960s, although it was not until the turn of the 21st century that ATMs became known for their role in human obesity-associated pathologies (56, 57). ATMs are critical to the healthy metabolic function of fat depots, orchestrating the production of mediators that induce mitobiogenesis, lipolysis, and mitochondrial uncoupling within adipocytes (56). Dysfunctional ATMs lead to the development of metabolic pathologies and beyond, including adipose fibrosis, chronic inflammation, lysosomal storage diseases, obesity, insulin resistance, diabetes, autoimmune diseases, and cancer (51, 56, 58–67). On the other hand, “alternatively activated” ATMs promote BeAT through type 2 cytokine release to improve metabolic function in the mouse (62, 68). Recent studies have shown that unhealthy maternal diets, including high fat and low protein, program offspring immune cell progenitors, ultimately impacting long-term immune memory (69–71). Maternal high-fat diet during lactation alters milk lipids resulting in proinflammatory ATMs in offspring, further suggesting a role of ATMs in the pathophysiology of metabolic programming of disease in early life (72). Importantly, the heightened sensitivity of ATMs to the environment, especially to lipids supplied during critical early-life windows, emphasizes the need to understand how ATM and AT develop in unison and offers great potential for the identification of novel intervention targets in the battle against obesity, diabetes, and metabolic dysfunction. Our review delves into the dynamic complexity of ATM diversity, their essential roles in AT, their potential role in early-life metabolic programming, and their overall impact on metabolism and health, illuminated by the latest technological advancements.

DISCUSSION

Early Life Metabolic Programming of AT

The “first one thousand days,” from conception to two years of age, is of particular developmental importance (73). Although health outcomes can be influenced before conception and beyond the first thousand days, recognizing this period as a critical “first hit” in metabolic programming has been instrumental. The importance of early-life events in establishing life-long susceptibility to obesity is now widely accepted. Adverse conditions during fetal development, for example, including maternal obesity, under- or overnutrition, gestational diabetes, fetal growth restriction, and exposure to certain chemicals, particularly endocrine disruptors, program persistent changes in feeding behavior and/or metabolism (74–80). Numerous studies have established that maternal obesity is associated with changes in the levels of adipokines, which are secreted from AT (81, 82). These adipokines, including adiponectin, interleukin 6 (IL-6), leptin, and tumor necrosis factor-alpha (TNF-α), provide a mechanistic link between maternal adiposity, fetal development, and future susceptibility to obesity. Likewise, perinatal exposure to endocrine-disrupting chemicals is another important mechanism that establishes persistent susceptibility to obesity (83–85). A recent study using a lineage tracing paradigm suggests that adipocyte progenitors that give rise to adult adipocytes in mice are established during fetal development, highlighting the importance of this developmental window for later life metabolism (86). In addition to the fetal window, metabolic development is shaped by adverse early postnatal conditions and exposures including postnatal maternal obesity, maternal overnutrition, postpartum diabetes, “catch-up” growth, reduced frequency and duration of breastfeeding, breast milk composition, formula feeding, and early introduction of solid foods (72, 75, 77, 79, 87–92). These and many other observations have led to the current view that developmental programming in early life sets the stage for later life obesity and that the perinatal period is a critical developmental window (Fig. 1). After this period, dietary and lifestyle choices act additively, further shaping long-term health trajectories.

Figure 1.

During the critical window of development from conception to 2 yr, known as the “first thousand days,” developmental conditions such as gestational diabetes, maternal obesity, malnutrition, exposure to endocrine disruptors, reduced frequency and duration of breastfeeding, and formula feeding lead to increased obesity risk in offspring. Nutrient exposures from fetal circulation and breastmilk, including carbohydrates (Carbs), amino acids (AA), fatty acids (FAs), and breastmilk-specific alkylglycerols (AKG) program adipose tissue macrophage (ATM) progenitors [hematopoietic stem cell progenitors (Pro-HSC) and erythromyeloid progenitors (EMP)]. ATMs subsequently choreograph adipose tissue development and function, imparting lifelong metabolic consequences (23–26, 56, 69–71, 74–80, 87–98). (Generated in BioRender).

Human AT development is stimulated, in part, by maternal lipid substrates and signals, and begins in the third trimester (99–101). In mice, the interscapular BAT depot develops during gestation, whereas other depots begin adipose expansion after parturition (101). As offspring transition from maternal lipids and carbohydrates supplied by blood in utero to enteric absorption during breastfeeding, the lipid-laden milk diet provides fuel for both thermogenesis and adipogenesis (102, 103). Infants efficiently absorb the diverse spectrum of lipids present in breast milk, the composition of which is reflected in infant plasma (99, 104–108). Development of healthy AT early in life is a critical determinant of lifelong metabolic health (109–111). Healthy fetal and neonatal development requires a delicate balance between over- and undernutrition, with either condition resulting in developmental problems across numerous species (112–122). Pre- and postnatal nutrient composition can guide molecular interventions that may prevent or reverse metabolic disease risks. For example, thermogenic BeAT is very important for infant maintenance of body temperature, and prolonging BeAT was shown to protect against obesity and metabolic dysfunction (123, 124). Conversely, accelerated WAT expansion is seen in childhood obesity and has negative long-term health consequences (125). Resident WAT stem-like cells (ASCs) expressing platelet-derived growth factor receptor alpha (PDGFRα), commonly known as preadipocytes, are an embryonically derived progenitor population required for WAT development that can differentiate into BeAT and WAT, depending upon external stimuli (126, 127). Determination of adipocyte fate (BeAT vs. WAT) is regulated primarily by alanine serine cysteine transporter-1 (Asc1) (128), and by PDGFRβ, which inhibits BeAT development (129). Factors influencing not only accelerated adipose accumulation but also the type of adipose formed, are important when considering the nutritional constituents promoting healthy early-life metabolism.

Our laboratory and others have shown that the breast milk composition is critical to infant adipose development and subsequent metabolic health outcomes, especially with regards to n3 (eicosapentaenoic acid, EPA; docosahexaenoic acid, DHA) and n6 (arachidonic acid; AA) PUFAs and their metabolites (87, 93, 94, 123, 130, 131). Dietary sources of n3 FAs include fish oil, flax seeds, and walnuts, whereas sources of n6 FAs include sunflower, safflower, corn, and peanut oils (132, 133). The Western style diet has led to drastically increased consumption of n6 FAs due to the prevalence of vegetable oils (133). Maternal intake of FAs is reflected in the breast milk of lactating mothers and directly impact offspring development (88, 89, 134), though formula feeding does not provide equivalent FA content (135). We found that increases in the breast milk ratio of AA to EPA + DHA predicted the increase of infant WAT accumulation over the first four months of life (87). Conversely, our studies using rodent models show that low ratios of AA to EPA + DHA during fetal and postnatal development protected against adult diet-induced obesity, improved glucose clearance, and associated with decreased methylation of the peroxisome proliferator-activated receptor (PPARγ2) proximal promoter, enhancing BeAT development (93). Separate studies in rats demonstrated similar results, with administration of n3 FA before and during gestation and lactation resulting in reduced insulin resistance (136). Others have determined that exposure to cold induces BAT production of the lipokine 12,13-dihydroxy-9Z-octadecenoic acid (12,13-diHOME), which promotes FA uptake in metabolic tissues (137). Isganaitis and coworkers (131) identified 12,13-diHOME and other thermogenic metabolites, including succinate, 9,10-epoxy-9(Z)-octadecenoic acid (9,10-epOME), 12,13-epOME, and lyso-phosphatidylglycerol 18:0 present in human breast milk, which are associated with more moderate infant adipose accumulation. In mice, maternal obesity attenuates β-adrenergic signaling, impairs adaptive thermogenesis, and downregulates BAT markers [Ucp1, cell death activator CIDE-A (Cidea), peroxisome proliferator-activated receptor gamma coactivator 1-alpha (Pgc1α), and PR domain zinc finger protein 16 (Prdm16)] (95). Altogether, a detailed picture of important early-life stimuli associated with the development of obesity in offspring is beginning to emerge and additional factors are covered in several recent reviews (5, 56, 138–144).

Optimal programming of ATMs during neonatal and perinatal stages is essential (56), given their embryonic origin (55, 145–147), and their role in establishing and maintaining thermogenic, metabolically healthy AT (96). ATMs ensure proper development and maintenance of AT by clearing dead adipocytes and dysfunctional mitochondria, and secreting signals that support BeAT function. For example, breast milk-specific lipids known as alkylglycerols are metabolized by ATMs into platelet-activating factor (PAF), which induces BeAT through IL-6/signal transducer and activator of transcription (STAT3)-mediated transcriptional induction of transmembrane protein 26 (Tmem26), Ucp1, cytochrome c oxidase subunit 7A1 (Cox7a1), iodothyronine deiodinase 2 (Dio2), Cidea, and Pgc1α (96). Cold exposure induces IL-4 secretion by eosinophils, which stimulates alternative ATM activation and catecholamine production (62, 148). Early studies using mice with impaired type 2 immune responses suggested these catecholamines promoted BeAT development by enhancing the expression of thermogenic genes such as Pgc1a, Asc1, and Ucp1 (148). However, the deletion of tyrosine hydroxylase, an essential enzyme in catecholamine synthesis, in hematopoietic cells had no effect on cold-induced thermogenesis (149), leading the authors to conclude instead that sympathetic activation plays an essential role in regulating AT lipolysis and thermogenesis. Another important mediator of adipose homeostasis is the IL-4-induced appetite regulator, neuropeptide FF (NPFF), which decreases obesity. NPFF binds ATM receptors (NPFFR2) to stimulate proliferation and alternative ATM activation genes [arginase (Arg1), IL-4ra, IL-10, and alkylglycerol monooxygenase (Agmo)] (150). Potential mechanisms through which ATMs mediate AT FA responses are depicted in Figs. 2 and 3. ATM exposure to n3 PUFAs induces the release of proresolving mediators, resulting in enhanced adipocyte lipid metabolism and improved metabolic health (Fig. 2) (151–154). Conversely, exposure to n6 PUFAs induces ATM inflammatory cytokine production, which promotes white adipogenesis, and chronic exposure results in increased meta-inflammation, insulin resistance, and obesity risk (Fig. 3) (87, 93, 123, 152, 153, 155).

Figure 2.

Omega-3 fatty acids (FAs) support production of pro-resolving mediators by adipose tissue macrophages (ATM), resulting in enhanced oxidative phosphorylation (OXPHOS) and FA β-oxidation. These pro-resolving mediators stimulate the beige adipose tissue (BeAT) phenotype by increasing lipolysis and decreasing lipid storage. The result is reduced inflammation, increased insulin sensitivity, and energy expenditure, and reduced obesity (151–154). (Generated in BioRender).

Figure 3.

Omega-6 fatty acids (FA) trigger adipose tissue macrophage (ATM) proinflammatory cytokines production, which stimulates adipose tissue stem-like cells (ASCs) for white adipogenesis and obesity predisposition (87, 93, 123, 152, 153, 155). (Generated in BioRender).

ATM Ontogeny and Diversity

Macrophages are a heterogeneous population of mononuclear phagocytes that are found throughout the body and function in both immunity and tissue homeostasis. This population includes both circulating monocyte-derived macrophages and TRMs. Previously, it was thought that all macrophages originated from bone-marrow-derived monocytes, which arose from hematopoietic stem cell (HSC) progenitors and were replenished throughout life. However, recent work has revealed that most TRMs originate embryonically and are long-lived within tissues (146). Fate-mapping studies have determined that multiple TRM populations are derived from HSC-independent embryonic precursors that originate from the fetal liver and yolk sac and are maintained by self-renewal (145). Early TRM progenitors are derived from late erythro-myeloid progenitors (EMPs), which serve as a developmental source of hematopoiesis before permanent blood system development and are formed in the extra-embryonic yolk sac beginning at mouse embryonic age 7 (E7) and human gestation week 3, depicted in Fig. 4 (147). By E8.5/E9, blood circulation is established, and by E9.5/E10, yolk-sac derived EMPs give rise to the first macrophages (yolk-sac macrophages, YSMs), without progressing through an intermediate monocyte stage. EMPs circulating from E8.5 to E10 infiltrate various tissues and establish what becomes the microglia of the brain, which persist throughout adulthood. EMPs also give rise to mast cell, granulocyte/macrophage, and erythroid progenitors that migrate to the fetal liver upon onset of blood circulation and establish lineages of basophils, mast cells, neutrophils, erythroid cells, monocytes, and macrophages. So-called “late EMPs” give rise to TRMs and fetal liver-derived monocytes, which colonize all tissues except the brain around E13.5 (146).

Figure 4.

Timeline of adipose tissue macrophage (ATM) fetal progenitor development and subsequent fates resulting in specific ATM phenotypes. The early wave of erythromyeloid progenitors (EMP) emerges from the yolk sac at mouse embryonic day 7.5 (E7.5) and gives rise to yolk-sac macrophages (YSM). The late wave of EMP appears at E8.5 and seeds the fetal liver (FL) with monocytes. At E10.5, hematopoietic stem cell progenitors (Pro-HSC) emerge from the aorta–gonad–mesonephros region and by E12.5, they differentiate into FL HSC, which serves as a source for bone-marrow HSC and later blood monocytes. Within AT, two major ATM phenotypes have been well-characterized: perivascular (PVMs) and lipid-associated macrophages (LAMs). PVM are mainly yolk-sac macrophage (YSM) derived and are maintained by self-renewal, with a small contribution from infiltrating monocytes. LAMs are mainly monocyte-derived, with some evidence of PVM to LAM differentiation. Although PVM and LAM are the best-characterized ATM phenotypes, a wide-variety of phenotypes have been documented and are recorded in Table 2 (28, 29, 32, 34, 35, 37, 54, 156–162). (Generated in BioRender).

HSCs first appear at E10.5, or gestation week 5 in humans, and originate from hemogenic endothelial cells of the hemogenic endothelium of the aorta. They mature in the fetal liver before traveling to the bone marrow and spleen (146). Studies suggest that adult-like HSC hematopoiesis begins in the bone marrow several days after birth, giving rise to circulating neonatal monocytes that replace the fetal-monocyte-derived macrophages of the aorta to establish a population of adult self-renewing macrophages (163). Although TRMs originate from EMPs, within tissues of the gut, dermis, serous cavities, and spleen, they are maintained, at least in part, by circulating monocytes of bone marrow origin, although this mechanism is poorly understood (146). Upon high-fat diet (HFD) challenge, monocytes are recruited into AT and differentiate into macrophages; however, the longevity of these recruited macrophages has not been established (164). For example, in studies of adult mice, crown-like structures (CLS) in epididymal WAT (eWAT) persisted as long as 6 mo after weight loss following obesity, and though the overall ATM number decreased, the proinflammatory profile of cluster of differentiation 11c positive (CD11c+) ATMs persisted (165). A summary of our current understanding of ATM ontogeny is depicted in Fig. 4.

Historically, macrophages have been divided into two functional states with a handful of subclassifications: proinflammatory, so-called “classically activated” macrophages (M1) and anti-inflammatory, so-called “alternatively activated” (M2) macrophages. This classification was supported by evidence seen in the polarization of macrophages as a product of two antagonistic pathways of arginine metabolism. “M1” polarization occurred as a result of the production of nitric oxide and citrulline in the inducible nitric oxide synthase (iNOS) pathway, whereas “M2” polarization resulted from the production of urea and ornithine in the arginase pathway (143). A thorough outline of the stimuli, resulting polarization, surface receptors, and subsequent gene induction, cytokine and chemokine secretion, and function of these older classifications of macrophages can be found in Table 1. Recent studies have revealed that macrophage phenotypes are much more plastic than suggested by this broad classification scheme, with a variety of phenotypic presentations unique to specific tissues. A wide range of phenotypes allows macrophages to respond to diverse environmental stimuli, including the milieu of signals that are present within tissues at any given time. The ability of ATMs to integrate stimuli from the adipose environment and choreograph tissue remodeling is critical to the maintenance of healthy metabolic function, such as clearance of dying adipocytes. Until recently, little was known about the tissue-specific macrophage populations within AT, much less the aberrant physiological mechanisms that contribute to metabolic dysfunction. For the purposes of this review, when referencing studies of historical significance, traditional classifications will be used so as not to make assumptions about the molecular identity of ATMs without significant evidence.

Table 1.

Characterization and functions of historical macrophage polarization states (143, 166)

Classification	Stimuli	Surface Receptors	Regulatory Transcription Factors	Cytokines/Chemokines Secreted	Functions
M1	LPS, IFN-γ, TNF-α, PAMPs, GM-CSF	CD80, CD86, CD68 TLR-2, TLR-4, iNOS, MHCII, CD11c, CCR7	NF-κB, STAT1, STAT5, IRF3, IRF5, HIF-1α, SOCS3	TNF-α, IL-1α, IL-1β, IL-6, IL-12, IL-23, CXCL9, CXCL10, CXCL11, CXCL16, CCL5	Proinflammatory, tissue injury, microbicidal, and tumoricidal.
M2	IL-4, IL-13, IL-10, IL-21, PPARγ Agonists	CD163, CD206, CD209, FIZZ1, YM1/2, CD301, Dectin-1, CXCR1, CXCR2	STAT3, STAT6, IRF4, KLF4, JMJD3, PPARδ, PPARα/β/γ, cMaf, cMyc	IL-10, TGF-β, CCL1, CCL17, CCL18, CCL22, CCL24, CXCL13, VEGF	Anti-inflammatory, angiogenesis, tissue repair, increased endocytic capacity, tumor formation/progression.
M2a	IL-4, IL-13	YM1/2, FIZZ1, Arg-1, CD206, IL-1R	–	TNF-α, IL-1α, IL-1β, IL-6, IL-12, IL-23, CXCL9, CXCL10, CXCL11, CXCL16, CCL5	Promote cell growth, tissue repair, enhanced endocytic activity.
M2b	Immune complex, TLR Ligands, IL-1β	IL-10R, IL-12R, CD86, IL-6R	–	TNF-α, IL-1β, IL-6, IL-10, CCL1	Promote Th2 differentiation, regulate immune response to parasitic, fungal, and bacterial infections.
M2c	Glucocorticoids, IL-10, TGF-β	CD163, CD206, TLR-1, TLR-8, Arg-1	–	IL-10, TGF-β, CCL16, CCL18, CXCL13	Phagocytosis of apoptotic cells.
M2d	TLR Antagonists	IL-10R, IL-12R	–	IL-10, VEGF	Promote tumor progression, promote angiogenesis.

CCL, CCR-like protein; CXCL, C-X-C motif chemokine; GM-CSF, granulocyte-macrophage colony-stimulating factor; HIF-1α, hypoxia inducible factor-1α; iNOS, inducible nitric oxide synthase; IRF, interferon regulatory factor; MHC, major histocompatibility complex; PAMP, pathogen associated molecular pattern; PPARγ, peroxisome proliferator-activated receptor gamma; SOCS3, suppressor of cytokine signaling 3; TGF-β, transforming growth factor beta; TLR, Toll-like receptor.

ATM Roles in Homeostasis and Inflammation

A number of studies have addressed the role of ATMs in homeostasis such as their adipose-specific lipid handling functions; however, much of this has been confined by the dated M1/M2 paradigm (167). ATMs from lean mice have low levels of metabolic activity, whereas HFD induces activation of catabolic pathways such as oxidative phosphorylation (OXPHOS) and glycolysis (168). OXPHOS and glycolysis both appear to contribute to proinflammatory cytokine release; however, glycolysis alone was responsible for increased IL-6 and C-X-C motif chemokine 1 (CXCL1) release from obese WAT, probably driven by increased hypoxia inducible factor-1α (HIF-1α) signaling (167). Phospholipids have been found to alter bioenergetics of bone marrow-derived monocytes and ATMs (169), with long-chain saturated FAs (SatFAs) inducing a proinflammatory state with increased lipid metabolism, glycolysis, and OXPHOS (170). On the other hand, anti-inflammatory effects are seen by exposure of ATMs to unsaturated FAs (170). Palmitate has been demonstrated to be proinflammatory, causing increased HIF-1α expression and glycolysis, resulting in IL-1β release (171), a mechanism that is reliant upon Toll-like receptor 4 (TLR4)-dependent bone marrow-derived monocyte priming by lipopolysaccharides (LPS) (172). For a better understanding of the effects of specific lipids on ATM phenotype, further studies are needed to characterize the handling of biologically relevant lipids, consistent with each of these more recently defined ATM types.

Significant cross talk occurs in AT between ATMs and adipocytes (40, 142, 173), specifically through ATM gene products such as fatty-acid-binding proteins [adipocyte protein 2 (ap2), fatty acid binding protein 4 (FABP4), and FABP5], and secretion of proinflammatory cytokines by classically activated macrophages (IL-6 and TNF-α) or anti-inflammatory cytokines by alternatively activated macrophages (IL-4, IL-10, IL-1B, etc.) (167, 173). The anti-inflammatory effects of ATMs are mediated by lipid catabolism master regulator PPARγ, and sequestosome 1 (p62), a scaffold protein that regulates signaling (32, 34, 174). Many targets of PPARγ are increased in obesity, including ATP binding cassette subfamily A member 1 (ABCA1), perilipin 2 (PLIN2), CD36, and IL-1B (174). PPARγ is secreted by adipocytes and induces monocyte differentiation into an alternatively activated, pro-resolving state. IL-4, IL-13, and IL-10 secretion by pro-resolving macrophages promotes insulin sensitivity and glucose homeostasis (166, 173). IL-10 promotes Janus kinase (JAK)/STAT3 signaling to prevent nuclear factor-kappa B (NF-κB) nuclear translocation and inflammatory cytokine production, while stimulating proinflammatory cytokine inhibitors (175). Acute IL-10 treatment improves global insulin sensitivity, and its expression is positively correlated with insulin sensitivity in humans, although hematopoietic deletion of IL-10 alone does not promote obesity or insulin resistance (176–179). Macrophage-derived IL-10, induced by insulin and LPS, suppresses hepatic glucose production and its impairment contributes to insulin resistance in obesity, highlighting IL-10’s therapeutic potential in regulating glucose homeostasis (180). IL-4 and -13 act through the interferon regulatory factor (IRF)/STAT pathway to activate STAT6 (166). IL-4 inhibits interferon gamma (IFN-γ) and production of other proinflammatory cytokines (175), and acts as a metabolic regulator by inhibiting adipogenesis through downregulation of PPARγ and CCAAT/enhancer-binding protein-alpha (CEBPα) (181). In addition, IL-4 promotes lipolysis through the cyclic adenosine 3,5-monophosphate (cAMP)/protein kinase A (PKA) pathway, leading to phosphorylation of hormone-sensitive lipase and perilipin, thereby reducing lipid storage and improving metabolic efficiency (181, 182). ATMs also regulate insulin sensitivity through exosomal transfer of miRNA, and treatment with exosomes derived from ATMs of lean mice improved insulin sensitivity in obese mice (183). On the other hand, secretion of TNF-α by hypertrophied adipocytes leads to the recruitment of classically activated macrophages. Activation of TLR4 by ligands such as SatFAs induces macrophage c-Jun N-terminal kinase (JNK) and NF-κB signaling pathways, promoting the secretion of TNF-α by proinflammatory macrophages. TNF-α stimulates adipocyte TNF receptors and promotes downstream activation of adipocyte JNK and NF-κB signaling pathways, inhibiting insulin receptor substrate-1 and 2, and ultimately leading to insulin resistance (166, 173).

Studies on apoptotic adipocyte clearing by ATMs have revealed a variety of lipid-handling mechanisms dependent upon the size of the droplets, including efferocytosis (<25 µm), fragmentation (25–50 µm), and CLS formation (>50 µm) (184). Adipocyte-ATM interaction occurs about 10 h after cell rupture, whereas CLS formation occurs around 24–48 h, and results in the upregulation of lipid metabolism, metabolic activation, and expression of inflammatory cytokine genes (184). A hallmark of adipocyte clearance from obese AT is the presence of CLS, which are composed of ATMs that have properties of both proinflammatory cells with the expression of IL-1β, IL-6, and TNF-α, and pro-resolving macrophages, with an increase in OXPHOS, again challenging the overly simplistic classical activation model and emphasizing the need for further studies (184).

Through the use of dietary interventions, mouse genetic models, and proteomics, Coats et al. (185) demonstrated that so-called “metabolically activated” ATMs perform adipose maintenance by clearing dead adipocytes via lysosomal synapse formation and release inflammatory cytokines under settings of prolonged HFD exposure. These physiological effects are coordinated through NADPH oxidase 2 (Nox2), Tlr2, Tlr4, and myeloid differentiation primary response 88 (Myd88) signaling, which result in differing outcomes, depending on the degree of activation (185). Upon release of free FAs following adipocyte apoptosis, ATMs increased the expression of genes associated with lipid metabolism (Plin2, Cd36, and Abca1), and increased inflammatory signaling resulted in CLS formation, both facilitating clearance of dead adipocytes. After long-term exposure to HFD, persistent inflammatory signaling leads to metabolic dysfunction and tissue damage (185). Separate studies have demonstrated that the ability of ATMs to respond to SatFAs depends upon Tlr4, Myd88, and TIR domain-containing adaptor protein-inducing interferon β (Trif) for HFD-induced myelopoiesis (specifically of ATMs expressing CD11c), expansion of macrophage progenitors, and expansion of pregranulocyte macrophage progenitors, respectively. Tlr4 was also shown to inhibit proliferation of CD11c− ATMs, as well as promote CD11c+ ATM recruitment (156, 186). Though the process of apoptotic cell clearance by TRMs is generally anti-inflammatory in nature, with induction of IL-10, transforming growth factor beta (TGF-β), CD206, and CD163, clearing of apoptotic adipocytes induces inflammation (48, 184). Chronic exposure to inflammation further propels adipocyte apoptosis, resulting in increased inflammatory cytokine production, meta-inflammation, impaired endocrine regulation of energy, impaired insulin sensitivity, and metabolic syndrome (64, 186).

AT inflammation is mitigated by n3 PUFAs via the G-protein-coupled receptor 120 (GPR120), also known as free fatty acid receptor 4 (FFAR4). GPR120 binds medium and long-chain PUFAs, and n3 PUFAs (187, 188). GPR120 is highly expressed in both adipocytes and ATMs, has been shown to be involved in BeAT metabolism and adipogenesis, and dysfunction of GPR120 is associated with obesity in mice and humans (189–193). Upon n3 PUFA stimulation in vivo, GPR120 induces anti-inflammatory effects by repressing TLR4-mediated inflammatory responses in ATMs. GPR120 complexes with β-arrestin 2 and sequesters transforming growth factor-β-activated kinase binding protein 1 (TAB1), inhibiting downstream phosphorylation of transforming growth factor-β activated kinase 1 (TAK1) and subsequent production of proinflammatory signaling molecules NF-κB and JNK and proinflammatory cytokines (IL-6 and TNF-α) (190). In vivo exposure to n3 PUFAs improved insulin sensitivity and hepatic lipid metabolism, and decreased hepatic steatosis, in a GPR120-dependent manner. Moreover, n3 PUFA activation of GPR120 resulted in a shift from the HFD-induced inflammatory ATM gene signature [TNF-α, IL-6, monocyte chemoattractant protein 1 (MCP-1, aka CCL2), IL-1β, iNOS, and CD11c] to that of anti-inflammatory ATMs [rodent-specific chitinase-like protein (Ym1), Arg1, IL-10, C-type lectin domain family 7, member A (Clec7a), and Clec10a] (190). Though the role of GPR120 in AT homeostasis has been investigated, there is evidence to suggest that the anti-inflammatory and metabolic effects of n3 PUFAs are not entirely dependent upon GPR120 (194), emphasizing the need for further studies.

The balance between proinflammatory and resolving ATM states is critical for AT development. Transient inflammatory signals within adipose are necessary for physiological processes such as AT development and remodeling in early postnatal life, whereas excessive inflammation results in pathological metabolic dysfunction (96). A population of CD44+, alternatively activated ATMs have been identified with elevated levels of 9-hydroxyoctadecadienoic acid (9-HODE) and 13-HODE, known ligands for PPARγ, as a consequence of upregulated arachidonate 15-lipoxygenase (ALOX15) expression (49). These ATMs induce adipogenesis through clearance of dying adipocytes, upregulation of ALOX15, release of 9-HODE and 13-HODE, and subsequent induction of PPARγ, resulting in de novo BeAT adipogenesis. There is clearly a fine line between homeostatic tissue remodeling and meta-inflammation within AT. In an obesogenic setting, hypertrophy of adipocytes and impaired lipolysis trigger apoptosis which, contrary to the clearance of cells in healthy AT, results in a proinflammatory state through release of proinflammatory cytokines (48, 184).

Cells with impaired mitochondrial function are capable of taking up functional mitochondria from circulation (195–204)—a process that often occurs in healthy AT and is driven by type two cytokine (IL-4 and IL-13) release from adipose immune cells (205, 206). In BAT, ATM clearance of mitochondria from secreted adipocyte extracellular vesicles is essential for efficient metabolic function and thermogenesis (207). Recently, Brestoff et al. (208) demonstrated the presence of a distinct subpopulation of ATMs, which participate in heparin sulfate-mediated mitochondrial uptake from surrounding adipocytes, serving a role in metabolic homeostasis. These ATMs have increased hypoxia and mitochondrial reactive oxygen species production, and reduced expression of mitochondrial homeostasis, collagen synthesis, and electron transport chain genes (208). Mitochondrial transfer is increased in CD206+/CD11c− ATMs compared with CD206−/CD11c+ ATMs (208). ATM/adipocyte mitochondrial transfer is inhibited in obesity (205, 209–214), likely driven by the decreased number of CD206+ ATMs and impaired transfer in both types of ATMs (208). Conversely, ATM/adipocyte lipid transfer via lipolysis and exosome transfer are increased in obesity and contribute to ATM differentiation (215). Overall, this results in meta-inflammation, oxidative damage, and impaired adipocyte function.

Novel Insights in the Era of Big Data

A number of revolutionary single-cell studies of adipose depots published over the past few years have provided a more thorough classification of ATMs, revealing a highly complex population of pro- and anti-inflammatory ATM phenotypes than had been documented historically (5, 6, 27–29, 34–39). As such, earlier studies that aimed to unearth the role of ATMs in the programming of AT, including investigations into early-life exposures, lack this nuanced identification of ATM phenotypes. Moreover, recent applications of these newer techniques are mostly limited to studies in adults and in the context of HFD challenge or obesity. Given the limited availability of early-life ATM data, recent ATM classifications are mainly derived from the adult setting. Future studies aimed at elucidating the role of ATMs in early-life AT should consider the wide range of ATM presentations covered below.

Resident ATMs from lean mice have been identified on the basis of expression of CD206 and differential expression of CD163 and T cell membrane protein 4 (Tim4), with Tim4+ macrophages being tissue resident and regulators of energy storage via PDGFcc, whereas CD163+/Tim4− macrophages are bone marrow monocyte-derived and C-C chemokine receptor type 2 (Ccr2) dependent (27, 216, 217). Further investigations of resident ATMs using genetic tracing models and unbiased single-cell mass cytometry (CyTOF) revealed 10 distinct ATM populations within lean eWAT, based on differential expression of CD206, Tim4, CD163, major histocompatibility complex class II (MhcII), CD11c, and lymphocyte antigen 6 family member C1 (Ly6c) (27). An impressive aspect of this study was the time course through which the investigation was carried, with eWAT sampled at juvenile ages of 2 and 5 wk, and adult ages of 8, 12, 16, and 23 wk. At the beginning of this time course, the predominant subpopulations within the eWAT were Tim4+ and Tim4+/CD163+ ATMs, which were MhcII−, and decreased into adulthood. Tim4+ and Tim4+/CD163+ macrophages, which were MhcII+, increased between 2 and 5 wk before gradually declining into adulthood. In contrast, Tim4−/CD163− and CD163+ cells, which were MhcII+, increased with age, and only a small number of CD206− macrophages were found in the lean state. Interestingly, lymphatic vessel endothelial hyaluronan receptor 1 (Lyve1) expression was represented in small portions of each cluster. Through the use of genetic models of Ccr2 and nuclear receptor subfamily 4, group A, member 1 (Nur77) deficiency and cell-fate mapping, it was confirmed that Ly6c^hi bone marrow-derived monocytes are major contributors to the CD206−, and the Tim4− ATMs of lean adipose, whereas the Tim4+/CD163+ and a small portion of Tim4+ macrophages are derived from YSMs or fetal liver monocytes (27). In addition, it was confirmed that Tim4+/CD163+ and CD11c+ macrophages are self-renewing, whereas CD163+ and Tim4−/CD163− macrophages are dependent on recruitment of circulating monocytes. During adult obesogenic challenge, the adipose ATM landscape undergoes remodeling, depending on the exact diet composition, with HFD resulting in increased inflammatory CD11c+ macrophages, and decreased MhcII expression in Tim4+ and Tim4+/CD163+ macrophages (27). More recently, techniques such as cellular indexing of transcriptomes and epitomes by sequencing (CITEseq), paired with single-cell RNA-sequencing (scRNA-seq) studies of weight cycling in adult male mice have corroborated that TRMs retain CD206 expression but lose CD163 expression with obesity, which was not reversed with weight loss or cycling. Furthermore, all ATMs, but especially the lipid-associated macrophage (LAM) population, became more inflammatory, which persisted through weight loss, with weight-cycled LAMs being the most inflammatory (218). Adipocyte energy storage and lipid synthesis mechanisms are thought to be regulated by ATM PDGFcc production by Tim4+ vWAT ATMs (217). Specifically, obesity-associated genes such as Pparα, isocitrate dehydrogenase 2 (Idh2), and insulin-induced gene 1 (Insig1) were regulated by PDGFcc.

A study on the effects of a 12-wk HFD on murine eWAT ATMs described the presence of three main macrophage subpopulations, with variable expression of CD11c: proinflammatory, lipid-laden Ly6c− CD9+ ATMs within CLS, a Ly6c− CD9− population of ATMs, and adipogenic Ly6c+ ATMs outside of the CLS (37). Upon HFD exposure, CD9+ ATMs and Ly6c+ ATMs, both derived from bone-marrow, accumulated within the eWAT, whereas Ly6c− CD9− ATMs did not. The CD9+ ATMs had a proinflammatory gene expression and chromatin landscape, whereas Ly6c+ ATMs were associated with physiological adipose function, confirmed using assay for transposase-accessible chromatin using sequencing (ATAC-seq), scRNA-seq, and bulk RNA-seq (37). Further studies revealed that murine adipose remodeling induced by three days of β3 adrenergic receptor activation at 8 to 9 wk of age resulted in an alternatively activated, osteopontin-dependent macrophage response that promoted both macrophage expansion, and ASC1a progenitor recruitment to CLS, giving rise to an adipogenic niche (36). This study described six different subpopulations of ATMs; most notably, ATMs associated with CLS were enriched in the genes triggering receptor expressed on myeloid cells 2 (Trem2), a lipid receptor, and CD9, aligning nicely with other studies (28, 29, 34, 36, 157, 218). Another scRNA-seq study investigated ATMs in the context of adult HFD-induced obesity in mice and corroborated the presence of three distinct macrophage populations within WAT, with a conserved population of LAMs that express Trem2, expression of which was exclusive to LAMs (34). The LAM population was further characterized by expression of genes associated with phagocytosis and lipid metabolism. These findings confirm the assertion that CD9+/CD63+ ATMs are, in fact, the LAM population, as well as LAM functional dependence on Trem2. In addition to the LAMs, a population with a gene expression signature attributed to perivascular macrophages (PVMs) was described with high expression of resistin-like alpha (Retnla), CD163, Lyve1, and CD209f (34). A study from earlier the same year, characterized “interstitial” macrophages of 5- to 10-wk-old mice as two distinct populations, with a Lyve1^loMhcII^hiCx3cr1^hi population residing alongside the nerves and being of blood monocyte-derived origin, and a Lyve1^hiMhcII^loCx3cr1^lo population, which also expressed CD206, associated with blood vessels—likely the PVMs described earlier (38).

Other studies of ATMs in eWAT of adult HFD-induced obese mice describe four distinct types of ATMs, characterized as CD11b+/CD206^lo/MhcII^hi/CD64+/CD11c− [Pre-vasculature-associated macrophages (VAMs)], CD11b+/CD206^hi/MhcII^hi/Tim4^int (VAM1), CD11b+/CD206^hi/MhcII^int/Tim4^hi (VAM2), and CD11b+/CD206^int/MhcII^hi/CD64+/CD11c+ (Double positive macrophages) (32). Under normal dietary conditions, the highly endocytic, blood-monocyte-independent, vasculature-associated macrophages (VAMs) surround adipocytes and are associated with vasculature, where they self-renew, orchestrate the establishment of healthy adipose, maintain adipose homeostasis, and serve as an endocytic defense mechanism against infection. These populations likely encompass PVMs, particularly the Tim4^hi VAM2 population, given the tissue-resident status of Tim4+ macrophages (27, 216). Upon HFD exposure, a dramatic increase was seen in the monocyte-derived double-positive macrophages, which expressed genes with an anti-inflammatory signature, and the endocytic activity of the VAMs was impaired. Silva et al. (32) concluded that double-positive macrophages are similar to the CD9+ LAM population described earlier, which are of mixed origin and form CLS to mitigate HFD effects in AT, along with VAMs. Another scRNA-seq analysis of eWAT ATM populations in an adult HFD mice paradigm recapitulated these findings in greater detail. Harasymowicz et al. (28) described five distinct macrophage populations, four of which were altered by HFD, and the fifth, expressing Fabp4, growth/differentiation factor 15 (Gdf15), and Cxcl12, was unaffected. Two inflammatory ATM populations, both expressing serum amyloid A 3 (Saa3) and Il6, were significantly increased in adult HFD conditions. The first was characterized as being the LAM population, responsible for CLS and expressing CD9, CD11c, matrix metalloproteinase-12 (Mmp12), lipase A (Lipa), CD36, and Trem2, whereas the second expressed proteoglycan 4 (Prg4), macrophage receptor with collagenous structure (Marco), secretory leukocyte protease inhibitor (Slpi), and Cxcl13. In agreement with other studies, two distinct ATM populations expressing markers of self-renewing VAMs such as CD206, MAF BZIP transcription factor B4 (Mafb4), CD163, Krüppel-like factor 4 (Klf4), Clec10a, and Lyve1 were characterized as expressing either Mmp9 or prostaglandin-endoperoxide synthase (Ptgs2) (28). Investigations into obesity-induced inflammation implicated Wnt/β-catenin signaling as an important driver of Saa3 inflammatory macrophage response (219).

An additional scRNA-seq study characterized six distinct ATM populations within vWAT in adult mice—described as major, phagocytic, activated, resident, stem-like, or heme- along with a population of macrophage-like B cells (35). This study used scRNA-seq to characterize immune populations of vWAT from control, 24-wk HFD obese, and 23-wk HFD mice with 2 wk of calorie restriction to serve as a weight-loss model. The major macrophage population accounted for nearly half of ATMs and was mainly of embryonic origin, with ∼95% tissue-resident, and increased expression of folate receptor beta (Folr2) and growth arrest-specific 6 (Gas6) (markers of resident cells). The obese AT showed an increase in total and major macrophages, and emergence of an obesity-specific macrophage population. This obesity-induced macrophage population was termed “phagocytic macrophages” and expressed markers of phagocytosis and endocytosis, including PYD and CARD domain (Pycard), Fc epsilon receptor 1g (Fcer1g), Fc-gamma receptor 4 (Fcgr4), platelet and endothelial cell adhesion molecule 1 (Pecam1), and Annexin 1 (Ax1), among others. Interestingly, ATM population proportions were partially restored to the lean state following the 2-wk calorie restriction for the major and stem-like macrophages; however, a unique signature of activated, resident, and phagocytic macrophages persisted. Using pseudo-time analysis, lean adipose was shown to have two pseudo-temporal bifurcations, with one branch giving rise to monocytes, activated and stem-like macrophages and macrophage-like B cells (65.3%), and the other branch consisting of the major and resident macrophages (major macrophage fate 1; 29.3%) (28). The obese adipose pseudo-temporal analysis revealed two additional branches, consisting of 56.3% of major macrophages (major macrophage fate 2) and 11.9% of phagocytic macrophages in the obese ATMs, with only 21.1% in the monocyte, stem-like, activated B cell category. The adipose from the calorie restricted group showed a similar shift away from monocytes (16.8%) to major macrophages (fate 2; 20.5%) but with a greater increase in phagocytic macrophages (57.5%) (35). It is possible that the phagocytic population described here overlaps with the LAMs described by others. Overall, these observations suggest that caloric-restriction partially reversed the obesity-associated ATM profile; however, it is also clear that an activated-macrophage imprint remained, with potential implications for future metabolic dysfunction.

In studies of human adipose, depot-specific alterations associated with obesity have also been characterized through scRNA-seq. One standout study profiled the human sWAT landscape in both lean and obese states and identified a total of 28 distinct cell types, three of which were ATM populations (29). Recapitulating mouse studies, a population of LAMs was found, which expressed lipoprotein lipase (LPL), CD9, and TREM2, and PVMs, which expressed complement component 1q (C1Q), LYVE1, and Selenoprotein P (SELENOP). In addition, an inflammatory macrophage (IM) was also distinguished by the expression of CXCL3, TNF, and CCL3L1 (29). As in mouse studies, the PVM population was decreased, whereas the LAM and IM populations accumulated in obese AT, and ligand-receptor interactome analysis confirmed their role in adipose inflammation. Pseudo-time analysis also confirmed that obesity-induced preferential differentiation of monocytes into IMs rather than PVMs via cytokine signaling (29). Additional scRNA-seq investigations into human vWAT and sWAT in the context of obesity also defined three distinct subtypes based on differential expression of CD206 and CD11C (157). Within vWAT, CD206+ and CD206+/CD11c+ ATMs correlated with metabolic syndrome and increased in obesity; though, the same was not seen in sWAT or vWAT CD11c+ ATMs. Ontogeny investigations suggest the CD11c+ and CD206+/CD11c+ ATMs are derived from blood monocytes and are proinflammatory, whereas the CD206+ have an anti-inflammatory profile (157). Another analysis of obese vWAT and sWAT depots revealed five distinct CD68+ macrophage clusters, which displayed similar properties to those seen within mice (39). A population of CD9+ metabolically active ATMs expressed CD36, LIPA, FABP4, and LPL, corresponding to the LAMs seen in mice. Moreover, a population of inflammatory ATMs was observed with increased expression of CXCL2, CXCL3, CXCL8, CCL3, and IL1B, and “M2”-like ATMs expressing FOLR2 and KLF4 (39). Given the difficulty in obtaining human samples, there is a great need for more extensive characterization of the human AT landscape.

Though the wide spectrum of ATM phenotypes necessitates further research, multiple studies have demonstrated the presence of at least two major ATM populations that play a role in homeostatic AT maintenance. A summary of these various ATM descriptions can be found in Table 2. LAMs are responsible for efferocytosis of adipocytes, which results in the formation of CLS within AT (28, 29, 32, 34, 37, 217). Although LAMs have been well documented, there is still disagreement over the proportion of LAMs that are EMP-derived versus HSC-derived. There is general agreement with respect to the dual anti- and proinflammatory nature of LAMs, further solidifying the complexity of ATMs. The second major ATM population, the PVMs, are EMP-derived, self-renewing, and involved in AT homeostasis under lean state conditions, with both metabolic and immune functions. Although most researchers agree that there is at least one other major macrophage phenotype, often characterized as having inflammatory properties and being of HSC origin, not enough evidence has been unearthed to fully characterize this population. Some researchers have characterized as many as 10 distinct ATM populations (27); however, a substantial amount of work remains to be done unveiling the true spectrum of ATM identities, especially with regards to early life programming.

Table 2.

Recent advances in characterization of adipose tissue macrophages, including lipid-associated and perivascular macrophages, as well as other phenotypes documented across literature (27–29, 32, 34, 35, 37, 39, 157, 218)

	Species	Lineage	Markers	Gene Expression	Functions	References
PVM	Mouse	–	CD163+, CD209f+, Lyve1+, Retnla+	Cbr2, F13a1, C4b	–	(34)
	Mouse	EMP	CD11b+, CD206HIGH, MhcIIHIGH, Tim4INT, CD163+, CD209, Retnla+	Ifnb1	AT establishment and maintenance, endocytosis, vascular immune function	(32)
	Mouse	EMP	CD11b+, CD206HIGH, MhcIIINT, Tim4HIGH, CD163+, CD209, Retnla+	Cltc, Clta, Snx5, Snx2, Ap2a2, Eps15, Ap2b1	AT establishment and maintenance, endocytosis, vascular immune function	(32)
	Mouse	EMP	Lyve1+	Folr2, Gas6, Klf2, Pf4	–	(35)
	Mouse	EMP	CD206+, CD163+, Lyve1+	Mafb4, Klf4, Clec10a, Mmp9	ECM organization	(28)
	Mouse	EMP	CD206+, CD163+, Lyve1+	Mafb4, Klf4, Clec10a, Ptgs2	COX-2 pathway enrichment	(28)
	Mouse	EMP	CD206+, Tim4+, CD163+, MhcII+	–	–	(27)
	Mouse	EMP	CD206+, Tim4+, CD163+	–	–	(27)
	Mouse	EMP	CD206+, Tim4+, MhcII+	–	–	(27)
	Mouse	EMP	CD206+, Tim4+	–	–	(27)
	Mouse	EMP	CD206+, CD163+	Klf4, Stab1, Cbr2	–	(218)
	Human	–	CD206+, LYVE1+	SELENOP, C1Q, VEGFA, TGFβ, PDGFC	Homeostasis	(29)
	Human	–	CD206+	CD36, MARCO, MERTK, STAB1, CD209, CD163L1, TIMD4	Antigen presentation, scavenging	(157)
LAM	Mouse	HSC	CD11b+, Ly6c−, CD9+, CD63+	Lpl, Plin2, Lamp2, Acp5, Ctss, Ccl2, Il1a, Il18, Tnf	Lysosomal function, CLS formation, efferocytosis, proinflammatory cytokine production	(37)
	Mouse	HSC	Trem2+, CD9+, CD63+, CD68+, CD36+	Nceh1, C1qa, Spp1, Lipa, Lpl, Ctsb, Ctsl, Fabp4, Fabp5, Lgals1, Lgals3	CLS formation, efferocytosis	(34)
	Mouse	EMP > HSC	Trem2+	Folr2, Gas6, Apoe, Sepp1, Cd74, H2-Aa, H2-Eb1	Lysosomal function, chemokine activity, antigen processing and presentation	(35)
	Mouse	HSC	CD9+, CD11c+, Trem2+	Saa3, Il6, Mmp12, Lipa, Cd36	CLS formation, efferocytosis, lipid metabolism, phagocytosis	(28)
	Mouse	HSC	Trem2+, CD9+	Lpl, Rgs1	Lipid metabolism	(218)
	Mouse	HSC	CD11b+, CD206LOW, MhcIIHIGH, CD64+, CD11c+, CD9+	Il10, Tgfb1, Il1rn, Tnf, Cxcl1, Cd44	CLS formation, efferocytosis	(32)
	Human	–	CD9+, CD16HIGH, CD206HIGH	–	CLS formation, efferocytosis	(37)
	Human	–	TREM2+	TIMP1, TIMP3, ALDOA, APOC1	Lipid metabolism, phagocytosis, oxidative phosphorylation	(34)
	Human	–	CD9+	Lpl, LIPA, CD36, FABP4	Lipid metabolism	(39)
	Human	HSC	TREM2+, CD9+, CD206+, CD11c+	LPL, TNF, IL-18, IL-1β, CXCL8, PDGFβ	Inflammatory cytokine production, CLS formation, lipid metabolism	(29)
	Human	–	CD11c+, TREM2+, CD9+	OLR, Clec7A, LY75, LPL, SPP1	CLS formation, efferocytosis	(157)
	Human	–	CD206+, CD11c+, TREM2+, CD9+	LPL, OLR1, Clec7A, LY75, SPP1	CLS formation, efferocytosis	(157)
Other	Mouse	HSC	CD206+, MhcII+	–	–	(27)
	Mouse	–	–	Saa3, Slpi	Efferocytosis	(218)
	Human	–	CD9−, CD16LOW, CD206LOW	–	–	(37)
	Human	–	–	IL-1β, CCL3, CXCL8, CXCL3, CXCL2	Inflammatory cytokine production	(39)
	Human	–	–	FOLR2, KLF4	–	(39)
	Human	HSC	CD11c+	CCL3L1, CXCL3, TNF, IL-18, IL-1β, CXCL8, PDGFβ	Inflammatory cytokine production	(29)

AT, adipose tissue; CLS, crown-like structures; COX, cyclooxygenase; ECM, extracellular matrix; EMP, erythro-myeloid progenitor; HSC, hematopoietic stem cell; LAM, lipid-associated macrophage; PVM, perivascular macrophage.

CONCLUSIONS

The global obesity crisis has necessitated a focus on the origins of metabolic dysfunction as a means to identify novel therapies for preventing or mitigating disease development. Mounting evidence supports the early origins of obesity, pinpointing the “first thousand days” as a critical period in which metabolic and immune health are programmed. AT has emerged as a crucial metabolic organ that is especially sensitive to early-life exposures such as inflammatory n6 and anti-inflammatory n3 FA. The increasing prevalence of n6 FA in maternal diets, and their association with offspring obesity, supports the need for further research into the impacts these FAs have on metabolic programming of AT. Emerging studies highlight the role of ATM, distinguishing them as master regulators of AT development, maintenance, and function. Recent application of scRNA-seq and other advanced technologies has uncovered more extensive phenotypic and molecular diversity among ATM populations than was previously appreciated. The overlapping gene expression signatures of AT cell types make it challenging to interpret data from whole tissue studies, highlighting the importance of methods that distinguish ATM and adipocyte populations. Importantly, contemporary technology has identified specific ATM populations associated with metabolic dysfunction and obesity in adulthood, and those that perdured following weight loss, which might persistently alter adipogenesis and metabolic responses during later dietary exposures. Cross-referencing these various ATM populations derived from their metabolic context will certainly be important to the study of development.

Nevertheless, the roles of these newly characterized ATM types, particularly in the context of early-life programming of AT development, require future exploration. How might maternally supplied lipids, such as alkylglycerols and oxylipins, mediate the complex AT-ATM interplay and promote the development of metabolically active BeAT early in life? Although current research underscores the potential of ATMs as targets for addressing metabolic dysfunction and obesity, challenges remain in studying early-life human immunometabolism due to limited access to infant tissue samples. With this in mind, unraveling of the fascinating complexity of the adipose-ATM choreography during early-life development must rely on mouse models. The use of knockout models for genes of interest, including those listed in Table 2, would begin to reveal mechanisms involved in the developmental programming of ATMs by maternal FA intake. Importantly, data from any models that use genetic manipulation should be carefully interpreted as genetic mutations have been known to affect normal physiology, as well as metabolic pathologies. Moreover, the use of maternal n6 FA exposure models should be used to help characterize the impact of such exposures on offspring macrophage programming and metabolic development. Future rodent studies should prioritize neonatal research, sex-specific differences, and the mechanisms of ATM programming during development, which could pave the way for novel therapeutic strategies in our ongoing battle to fight the escalating obesity epidemic.

DATA AVAILABILITY

Data will be made available upon reasonable request.

GRANTS

M.C.R. and K.H. are supported by the Presbyterian Health Foundation Science Award, the TSET Oklahoma Center for Adult Stem Cell Research Award, and the College of Medicine Alumni Association Award. K.A.Z. is supported by the National Institute of Diabetes and Digestive and Kidney Diseases Grants NIH R01DK129255 and NIH K01DK119375-01A1. P.R.N. is supported by the National Heart, Lung, and Blood Institute Grants NIH R01HL156856 and NIH R01HL137799 and the American Heart Association Grants TPA97002.

DISCLOSURES

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

AUTHOR CONTRIBUTIONS

K.B.H. conceived and designed research; K.B.H. prepared figures; K.B.H. drafted manuscript; K.B.H., G.P.M., P.R.N., K.A.Z., and M.C.R. edited and revised manuscript; K.B.H., G.P.M., P.R.N., K.A.Z., and M.C.R. approved final version of manuscript.

GLOSSARY

9,10/12,13-epOME

9,10/12,13-Epoxy-9(Z)-octadecenoic acid

12,13-diHOME

12,13-Dihydoxy-9Z-octadecenoic acid

n3

Omega-3

n6

Omega-6

AA

Arachidonic acid

ABCA1

ATP binding cassette subfamily A member 1

Acp5

Acid phosphatase 5, tartrate resistant

Agmo

alkylglycerol monooxygenase

ALDOA

Aldolase, fructose-bisphosphate A

ALOX15

arachidonate 15-lipoxygenase

aP2

Adipocyte protein 2

Ap2a2, Ap2b1

Adaptor-related protein complex 2 subunit alpha 2 or beta 1

APOC1

Apolipoprotein C1

Arg1

Arginase

Asc1

Alanine serine cysteine transporter-1

AT

Adipose tissue

ATAC-seq

Assay for transposase-accessible chromatin using sequencing

ATM

Adipose tissue macrophages

Ax1

Annexin 1

BAT

Brown adipose tissue

BeAT

Beige adipose tissue

C1Q

Complement component 1q

C1qa

Complement C1q A chain

C4b

Complement C4B

Cbr2

Carbonyl reductase 2

CCL

CCR-like protein

Ccr2, 7

C-C chemokine receptor type 2 or 7

CD11, 36, 163, 206

Cluster of differentiation 11, 36, 163, 206, etc.

CEBPα

CCAAT/enhancer-binding protein-alpha

CIDEA

Cell death activator CIDE-A

CITE-seq

Cellular indexing of transcriptomes and epitomes by sequencing

Clec7a, Clec10a

C-type lectin domain family 7 or family 10, member A

CLS

Crown-like structures

Cltc, Clta

Clathrin heavy chain C or light chain A

cMAF

c-Maf proto-oncogene

cMyc

MYC proto-oncogene

Ctss, Ctsb, Ctsl

Cathepsin S, B, L

COX-2, Ptgs2

Cyclooxygenase 2, Prostaglandin-endoperoxide synthase 2

Cox7a1

cytochrome c oxidase subunit 7A1 IL-10R, IL-12 R, etc.

Cx3cr1

CX3C motif chemokine receptor 1

Cxcl

C-X-C motif chemokine

CXCR

C-X-C motif receptor

CYTOF

Cytometry by time of flight

DHA

Docosahexaenoic acid

Dio2

Iodothyronine deiodinase 2

DOHaD

Developmental origins of health and disease

ECM

Extracellular matrix

EMP

Erythro-myeloid progenitor

EPA

Eicosapentaenoic acid

Eps15

Epidermal growth factor receptor pathway substrate 15

eWAT

Epididymal white adipose tissue

F13a1

Coagulation factor XIII A chain

FA

Fatty acid

FABP4, FABP5

Fatty acid binding protein 4 or 5

Fcer1g

Fc epsilon receptor Ig

Fcgr4

Fc-gamma receptor 4

FFAR4, GPR120

Free fatty acid receptor 4, G-protein-coupled receptor 120

FIZZ1

Found in inflammatory zone 1

Folr2

Folate receptor beta

Gas6

Growth arrest-specific 6

Gdf15

Growth/differentiation factor 15

GM-CSF

Granulocyte-macrophage colony-stimulating factor

H2-aa, H2-eb1

Histocompatibility 2, class II antigen A, alpha or E beta

HFD

High-fat diet

HIF-1α

hypoxia inducible factor-1α

HODE

Hydroxy octadecadienoic acid

HSC

Hematopoietic stem cell

Idh2

isocitrate dehydrogenase 2

IFN-γ

Interferon-γ

Ifnb1

Interferon beta 1

iNOS

Inducible nitric oxide synthase

Insig1

insulin-induced gene 1

IL

Interleukin

IL-10, IL-12

IL-10 receptor, IL-12 receptor, etc.

IRF

Interferon regulatory factor

JAK

Janus kinase

JMJD3

Lysine demethylase 6B

JNK

c-Jun N-terminal kinase

Klf2, 4

Krüppel-like factor 2 or 4

LAM

Lipid-associated macrophage

Lamp2

Lysosomal associated membrane protein 2

Lgals1, Lgals3

Galectin 1 or 3

Lipa

Lipase A

LPL

Lipoprotein lipase

LPS

Lipopolysaccharide

Ly6c

Lymphocyte antigen 6 family member C1

LY75

Lymphocyte antigen 75

Lyve1

Lymphatic vessel endothelial hyaluronan receptor 1

M1

Classically activated macrophage

M2

Alternatively activated macrophage

Mafb4

MAF BZIP transcription factor B4

Marco

Macrophage receptor with collagenous structure

MerTK

MER proto-oncogene, tyrosine kinase

MCP-1

Monocyte chemoattractant protein 1; aka Ccl2

MhcII

Major histocompatibility complex class II

Mmp9, 12

Matrix metalloproteinase-9 or 12

Myd88

Myeloid differentiation primary response 88

Nceh1

Neutral cholesterol ester hydrolase 1

NF-κB

Nuclear factor-κB

Nox2

NADPH oxidase 2

NPFF

Neuropeptide FF

NPFFR2

Neuropeptide FF receptor 2

Nur77, Nr4A1

Nuclear receptor subfamily 4, group A, member 1

OLR, OLR1

Oxidized low-density lipoprotein receptor 1

OXPHOS

Oxidative phosphorylation

p62, SQSTM1

Sequestosome 1

PAF

Platelet-activating factor

PAMP

Pathogen associated molecular pattern

PECAM1

Platelet and endothelial cell adhesion molecule 1

PDGFRα, β, c

Platelet-derived growth factor receptor alpha, beta, or c

Pf4

Platelet factor 4

PGC1α

Peroxisome proliferator-activated receptor gamma coactivator 1-alpha

PLIN2

Perilipin 2

PPARα, γ, δ

Peroxisome proliferator-activated receptor alpha, gamma, or delta

PRDM16

PR domain zinc finger protein 16

PRG4

Proteoglycan 4

PUFA

Polyunsaturated fatty acids

PVM

Perivascular macrophage

Pycard

PYD and CARD domain

Retnla

Resistin-like alpha

Rgs1

Regulator of G protein signaling 1

Saa3

Serum amyloid A 3

SatFA

Saturated fatty acids

scRNA-seq

Single-cell RNA sequencing

SELENOP, Sepp1

Selenoprotein P

Slpi

Secretory leukocyte protease inhibitor

Snx2, Snx5

Sorting nexin 2, 5

SOCS3

Suppressor of cytokine signaling 3

Spp1

Secreted phosphoprotein 1

Stab1

Stabilin 1

STAT1, 3, 5, 6

signal transducer and activator of transcription 1, 3, 5, or 6

sWAT

Subcutaneous white adipose tissue

TAB1

Transforming growth factor-β-activated kinase binding protein 1

TAK1

Transforming growth factor-β activated kinase 1

TGF-Β

Transforming growth factor beta

Tgfb1

Transforming growth factor beta 1

Tim4

T-cell membrane protein 4

TimD4

T cell immunoglobulin and mucin domain containing 4

TIMP1, TIMP3

TIMP metallopeptidase inhibitor 1

Tlr1, 2, 4, 8

Toll-like receptor 1, 2, 4, or 8

Tmem26

Transmembrane protein 26

TNF-α

Tumor necrosis factor alpha

TREM2

Triggering receptor expressed on myeloid cells 2

Trif

TIR domain-containing adaptor protein inducing interferon beta

TRM

Tissue resident macrophage

UCP1

Uncoupling protein 1

VEGF

Vascular endothelial growth factor

VAM

Vasculature-associated adipose tissue macrophage

vWAT

Visceral white adipose tissue

WAT

White adipose tissue

Ym1

Rodent-specific chitinase-like protein

YSM

Yolk-sac macrophage