Obesity and its co-morbidities, including type 2 diabetes mellitus (T2DM), metabolic-associated fatty liver diseases, cardiovascular diseases, hypertension, and atherosclerosis, continue to pose significant global health challenges in modern sedentary lifestyles.^1,2 Obesity progresses from an imbalance between high caloric intake and low physical activity to excessive body fat accumulation. Chronic low-grade inflammation, particularly within visceral fat, marks the development of obesity, alongside the progression of insulin resistance.³ These factors have prompted an increase in studies evaluating adipose tissue (AT) biology, function, and regulation in health and obesity.

AT is an endocrine and immunological organ that preserves the energy balance by storing surplus energy as lipid triglyceride (TG). Nearly two decades have passed since the field of immunometabolism first highlighted the critical role of AT leukocytes in regulating AT function and homeostasis.⁴ A network of innate and adaptive immune cells, such as adipose tissue macrophages (ATMs), dendritic cells (DCs), natural killer cells, innate lymphoid immune cells, cluster of differentiation 4 (CD4)⁺ T cells, CD8⁺ T cells, and γδ T cells, exists uniquely and co-localizes within AT, suggesting active communication both within the network and with other cell types.^5-7 In obesity, the shift in AT leukocyte composition toward pro-inflammatory profiles underscores their significant roles in modulating AT inflammation.^8,9 In this light, the increasing number of ATMs that precedes the development of insulin resistance hints at their crucial function in driving AT inflammation and orchestrating metabolic changes in obesity.^4,10 Nevertheless, unraveling the biology, behavior, and function of ATMs in the context of obesity and from the perspective of immunology and metabolism research remains a dynamic challenge. Recent advances in technology, such as single-cell and single-nucleus RNA sequencing, have rapidly expanded our understanding, revealing diverse subpopulations of ATMs in lean versus obese states and suggesting that each distinct ATM type has specialized mechanisms.^11,12 In this review, we provide an overview of the current understanding of ATM biology and summarize recent characterizations of the diverse ATM subpopulations in obesity. Furthermore, we discuss the extended functions and potential mechanisms of ATMs as key players in obesity, emphasizing molecular targets for future prevention and therapeutics.

General characteristics of ATMs

Traditional immunological definitions of macrophages vary by tissue origin and are context dependent.¹³ Generally, macrophages are innate immune cells that phagocytose pathogens and clear debris through sterile inflammation and repair processes.¹⁴ They exhibit plasticity in response to tissue-specific stimuli.^11,15 Markers such as CD64 and MER proto-oncogene, tyrosine kinase (MerTK), along with the classic marker F4/80, have improved macrophage identification and functional observation in AT.^16,17 ATMs can be identified in lean mice by the phenotype F4/80⁺ CD64⁺ CD206⁺ CD301⁺ CD11c^–. In healthy AT, macrophages maintain homeostasis by scavenging adipocyte debris, performing efferocytosis, promoting pre-adipocyte survival, buffering lipids, and regulating adipocyte lipolysis.⁵ ATMs also secrete anti-inflammatory cytokines, micro-RNAs (miRNA), and neural factors that are crucial for promoting AT insulin sensitivity and modulating wound healing and angiogenesis.^18,19

The proportion of ATMs undergoes a dramatic increase in obese AT, rising from around 10% to up to 40% of leukocytes.⁴ Whereas resident lean ATMs are primarily anti-inflammatory, obese ATMs highly express pro-inflammatory markers, indicating their role in modulating AT inflammation.⁹ ATM classification initially followed the M1/M2 paradigm, describing M1 ATMs as pro-inflammatory and M2 ATMs as anti-inflammatory.⁸ CD11c serves as a useful marker for distinguishing between recruited ATMs and resident ATMs, though it alone cannot identify ATMs due to its high expression in DCs.^8,17 Several groups, including ours, have suggested that CD64⁺ CD11c⁺ denotes pro-inflammatory M1 ATMs, and CD64⁺ CD11c^– denotes anti-inflammatory M2 ATMs. However, that classification is now viewed as outdated and oversimplified because it fails to encompass the diversity of ATMs found in obesity.^5,20,21

Development of ATMs

Our understanding of ATMs has expanded significantly now that we can trace their developmental origins and thus highlight ontological distinctions among macrophage types. In lean AT, the resident ATM populations, like those in most peripheral tissues, are derived from yolk-sac progenitors and have the capacity for self-renewal through proliferation within AT.^22,23 Lineage tracing in mice has identified early mesodermal erythromyeloid progenitors as the origin of ATMs.²⁴ However, various origins contribute to the diversity of tissue resident macrophages, and those populations change with age.²⁵ Bone marrow (BM) monocytes are well-known to be a source of resident ATMs, and they gradually infiltrate and replace the yolk-sac-derived ATMs over time (Fig. 1).²⁵ In blood circulation, Ly6C⁺ monocytes differentiate into Ly6C^– monocytes in steady states. However, in response to inflammatory chemokines such as C-C motif chemokine ligand 2 (CCL2), for instance in obese conditions, Ly6C⁺ monocytes become activated and migrate to inflammation sites such as AT, where they further differentiate into ATMs.^26,27 According to this theory, C-C chemokine receptor type 2 (CCR2)^low Ly6C^– monocytes tend to differentiate into anti-inflammatory ATMs, whereas CCR2^hi Ly6C⁺ inflammatory monocytes preferentially differentiate into pro-inflammatory ATMs.²⁸ In addition, local proliferation of resident ATMs, followed by their polarization into distinct ATM subtypes, might further contribute to the diversity of ATM subtypes in obesity.^9,29

Dynamic properties of ATMs during obesity

In steady states, ATMs reside between adipocytes or along vascular structures in AT. They express anti-inflammatory cytokines such as interleukin 10 (IL-10) and catecholamines, which regulate adipocyte lipid metabolism and clear apoptotic cells through efferocytosis.⁹ In lean conditions, resident ATMs are primarily maintained by in situ proliferation rather than the infiltration of monocyte-derived macrophages.^30-32 Evidently, early feeding with a high-fat diet (HFD) triggers a rapid proliferation of ATMs in mice.^9,29 Concurrent with adipocyte enlargement during obesity progression, signals from adipocytes have been suggested to induce ATM accumulation around dead adipocytes, forming crown-like structures (CLS).³³ Monocyte trafficking studies suggest that obesity increases the recruitment of monocyte-derived macrophages to AT, primarily localizing at CLS.^9,34,35 In established obesity, local ATM proliferation occurs predominantly within CLS, contributing to their elevated numbers.²⁹ Thus, increases in both the infiltration and proliferation of ATMs are suggested to contribute to their elevated number in obese AT (Fig. 1).^30-32

ATMs at CLS exhibit a pro-inflammatory phenotype, with induced expression of CD11c, CCR2, Toll-like receptor 4 (TLR4), and CD9 surface proteins.^9,34,35 This phenotype is associated with ATM functions of phagocytosing dead adipocytes and excess lipid secretions, preventing excess free fatty acids (FFA) secretion into the circulation.⁹ The high lipid content inside ATMs has been implicated to alter their metabolism, shifting them toward a pro-inflammatory state, which is related to decreased cell egress and enhanced cell survival.^9,29 These phenomena might also drive the increased number of ATMs in obese AT (Fig. 1).^30-32

Various factors in obese AT, including FFA, cholesterol, glucose, insulin, and lipopolysaccharide (LPS), have been identified as unique inducers, which has led to the identification of metabolically activated macrophage (MMe) subsets in obesity.^20,36,37 In vitro, MMes can be mimicked using BM-derived macrophages exposed to FFA, high glucose, and insulin.³⁶ MMes exhibit unique pro-inflammatory cytokine profiles, overexpressed lipid receptors, and unique lysosomal surface proteins that distinguish them from classically activated M1 ATMs.^20,36,37 Their distinct properties suggest that they are responsible for specific regulations within the AT niche. Although understanding of the regulatory mechanisms and functionality of MMes in obesity remains incomplete, recent single-cell studies have identified unique definitive markers that characterize various subpopulations within this ATM subset, highlighting the complexity of ATMs in obesity and underscoring the need to revisit their classification and characterization.

Advances in single-cell studies have elucidated the diversity of ATMs in obesity. Single-cell RNA sequencing primarily focuses on isolating stromal vascular cells (SVCs) or CD45⁺ cells from AT, excluding adipocytes. Meanwhile, single-nucleus RNA sequencing techniques use a mild mechanical lysis process to isolate nuclei from both adipocytes and SVCs. Additionally, spatial transcriptomic and single-cell mass cytometry add spatial information to single-cell landscapes.³⁸ In this section, we discuss findings about ATM diversity in murine models down to the single-cell level, as summarized in Table 1.

Heterogeneity and adaptability of resident ATMs in murine models

In lean states, resident ATMs primarily express CD206 and CD163 markers, characteristic of alternatively activated M2 macrophages. Single-cell studies demonstrate that perivascular macrophages (PVMs), with the signature genes resistin like alpha (Retnla), Cd163, lymphatic vessel endothelial hyaluronan receptor 1 (Lyve1), and mannose receptor C-type 1 (Mrc1), dominate resident ATMs,^39-42 followed by non-PVMs expressing Cd74, Fc receptor like 2 (Fcrls), and eosinophil-associated, ribonuclease A family, member 2 (Ear2).³⁹ A recent single-cell mass cytometry analysis identified four CD206⁺ subsets and two CD206^– subsets of lean ATMs.¹² CD206⁺ ATMs can be further categorized based on their T-cell/transmembrane immunoglobulin and mucin domain containing 4 (TIM4) and CD163 expression, with additional distinctions in major histocompatibility complex II (MHC-II) expression. Meanwhile, CD206^– ATMs express CD11c and Ly6C, which are highly prevalent in HFD conditions. Genetic lineage tracing and inducible labeling techniques have revealed that TIM4^– CD163⁺, TIM4^– CD163^–, and CD206^– populations originate from the BM, whereas TIM4⁺ CD163⁺ cells actively proliferate from an embryonic origin. Interestingly, aging alters the proportion of these ATMs, reducing TIM4⁺ and TIM4⁺ CD163⁺ populations and increasing CD163⁺, TIM4^– CD163^–, Ly6C⁺, and CD11c⁺ populations. TIM4⁺ CD163⁺ ATMs demonstrate active scavenging in vivo, and all ATM subtypes perform functions such as phagocytosis, endocytosis, and antigen processing, supporting our general understanding of ATM functions.

Emergence of variety in ATMs during obesity in murine models

Investigations in obese mice have revealed the emergence of CD11c, CD9, and triggering receptor expressed on myeloid cells 2 (TREM2) expression in ATM populations, alongside pro-inflammatory profiles.^8,41,42 Over time, the CD11c-expressing ATMs in obesity exhibit metabolic activation that favors both glycolysis and oxidative phosphorylation, categorizing them as MMes with unique lysosomal markers that regulate inflammatory cytokine secretion via TLR2, NOX2, and MYD88.^39-41,43 However, variability in CD11c expression across the Ly6C⁺, Ly6C^– CD9^–, and Ly6C^– CD9⁺ subsets in a single-cell analysis challenges its role as a definitive obese ATM marker.⁴² Instead, induced CD9 expression correlates with high intracellular lipid content, lysosomal-dependent lipid metabolism, and inflammatory responses akin to MMes, despite the co-expression of CD206. Moreover, CD9-expressing ATMs are predominantly found at CLS and display heightened lysosomal activity linked to adipocyte debris clearance.⁴²

Amid uncertainties about precise MMe markers and functionality, recent single-cell studies have identified a new subset of ATMs termed lipid-associated macrophages (LAMs) and characterized by TREM2 expression.^40,41,44 These TREM2-expressing ATMs co-express high levels of CD9 and exhibit a lipid activation profile similar to CD11c⁺ ATMs, which are also found in liver and brain tissue macrophages.^40,41,45 As shown in a single-cell analysis, Cd9-mid expressing ATMs increase in obesity, but Cd9-high Trem2-expressing ATMs predominate in obese ATM populations.⁴¹ The Cd9-high Trem2 ATMs are implicated in lipid metabolism, phagocytosis, and MHC-II antigen presentation.⁴⁰ Immunostaining validates the transcriptomic profile by demonstrating the colocalization of TREM2⁺ ATMs with CD9 at CLS, highlighting their characteristics of high intracellular lipid content and inflammatory profiles. Induction of TREM2 expression in ATMs has been suggested to prevent adipocyte hypertrophy and worsening AT metabolic derangements by promoting CLS formation. Importantly, TREM2 expression is absent in other AT leukocytes, underscoring its specificity as a marker for LAM subsets and potentially enhancing the identification of obesity-induced ATMs.

Further categorization of ATMs at the single-cell level in obese AT has identified specialized subsets with distinct functions, including phagocytic, efferocytotic, cycling, stem-like, B-cell-like, collagen-expressing, and iron-associated ATMs (Table 1).^39,40,43 Their specific roles and contributions to regulatory mechanisms in obese AT require further investigation, but changes in these ATM subsets underscore their unique functions. For instance, the activation of phagocytic ATM clusters in obesity significantly increases after calorie restriction, whereas the accumulation of LAMs does not completely reverse. This observation suggests that ATMs might help to restore AT homeostasis through the clearance of excess lipids and dead cells during calorie restriction.^32,40,43

The recruitment and accumulation of obese ATMs from monocyte origins are known to depend on the CCR2/CCL2 axis.⁴⁶ Single-cell investigations have confirmed that the majority of obese ATM subsets originate from monocyte-derived precursors, although another study suggests a mixed origin involving both embryonic and monocyte-derived cells.^12,40,42 Furthermore, local differentiation of monocytes into macrophages within AT has been proposed. For example, the emergence of LAM subsets is characterized by the loss of Ly6c2, Ccr2, and lysozyme 2 (Lyz2) expression and the acquisition of Cd9, Cd68, and Cd63 expression.⁴¹ Understanding the regulatory mechanisms of obese ATM development requires further in vivo confirmation and could open new avenues for targeting ATM function in obesity.

Significant gaps persist in elucidating the diversity of human ATMs and their relevance to obesity and metabolic disorders. Although the accumulation of ATMs in obese humans is evident and shares similarities with observations in mice, there are also distinctions. Variations in metabolic status among obese subjects further complicate the evaluation of human ATMs, particularly at the single-cell level. In this section, we discuss and summarize current characterizations of ATMs in human AT samples to provide insights into their diversity (Table 2).^{38,41,44,47-49}

Identification of human ATMs

Flow cytometry analyses typically utilize CD11b and CD14, which are also expressed by monocyte-derived cells, as markers to identify human ATMs. In lean states, resident ATMs can be further characterized as CD16^– CD68⁺ CD163⁺ CD204⁺ CD206⁺, analogous to anti-inflammatory ATMs in mice. In obesity, human ATMs show heterogeneity, with mixed inflammatory phenotypes and distinct depot regulation. Unlike in obese mice, evaluations of obese human AT consistently report increased double positive CD206⁺ CD11c⁺ ATMs in subcutaneous AT (SAT), compared with visceral or omental AT (VAT). However, the proportion of pro-inflammatory CD206⁺ CD11c⁺ ATMs relative to anti-inflammatory CD206⁺ CD11c^– ATMs is higher in VAT than SAT and correlates with insulin resistance status. The concurrent expression of CD11c with CD206 and CD163 in obese human ATMs suggests a mixed M1/M2 phenotype, indicating a distinct differentiation mechanism in resident human ATMs that leads to the upregulation of the CCR2, CD44, human leukocyte antigen–DR isotype (HLA-DR), CD40, CD38, CD274, and CD319 surface markers that are associated with a pro-inflammatory phenotype.⁵ Additionally, other reports indicate the emergence of pro-inflammatory “MMe” ATM subsets in obese human SAT and VAT, which can be distinguished from classic M1 ATMs by unique markers such as ATP binding cassette subfamily A member 1 (ABCA1), CD36, and perilipin 2 (PLIN2).^20,36,47

Single-cell resolution of human ATMs

Single-cell studies of human AT have identified several clusters of resident ATMs in steady states resembling M2-type ATMs, each characterized by unique expressions of genes such as KLF transcription factor 4 (KLF4), MRC1, and CD163 (Table 2). Although these subsets are suggested to have anti-inflammatory properties, their specific functional differences remain unclear.³⁸ The existence of PVM subsets has also been detected in human SAT, constituting the largest proportion, with CD206⁺ CD11c^– CD9^– surface markers and signature genes LYVE1, selenoprotein P (SELENOP), and C1Q.⁴⁷ Another study consistently described CD206⁺ CD11c^– ATMs as resident, reporting that they were enriched in scavenger receptor genes and showed low lipid staining. The percentage of PVM decreases in obesity, whereas that of the pro-inflammatory LAM (CD206⁺ CD11c⁺ CD9⁺) and inflammatory macrophage (IM; CD206^– CD11c⁺ CD9^–) subsets increases. Interestingly, although the PVM frequency remains highest even in overweight and obese conditions, these cells do not exhibit the ability to produce IL-1β or tumor necrosis factor α (TNF-α).⁴⁷

A trajectory analysis has shown that PVMs differentiate from classical monocyte-1 (Mo-1) subset in AT under the influence of anti-inflammatory cytokines (IL-1RA, IL-13, and IL-37) and transcription factors (Nanog homeobox [NANOG], MAF bZIP transcription factor B [MAFB], myocyte enhancer factor 2C [MEF2C], and GATA binding protein 4 [GATA4]). However, in obese conditions, PVMs can differentiate into LAMs that are regulated by pro-inflammatory cytokines, JUN-FOS signaling pathways, and hypoxic signaling pathways.⁴⁷ Previous studies have observed an increase in CD206⁺ CD11c^– resident ATMs in the VAT of obese diabetic patients, compared with lean and obese non-diabetic subjects.³⁷ PVM subsets are suggested to produce chemokines, such as CCL3, C-C motif chemokine ligand 3 like 1 (CCL3L1), CCL4, and CCL2, in obese AT, which induces the recruitment of myeloid cells and potentially contributes indirectly to the progression of AT inflammation.⁴⁷ Therefore, it remains unclear whether the observed changes in PVM numbers are a result of a homeostatic response or contribute to metabolically harmful obesity phenotypes.

Distinct from murine studies, LAM subsets are present in both lean and obese human AT.⁵⁰ This feature highlights the independence of human LAMs from obesity, allowing them to reside in AT regardless of obesity status. Furthermore, a higher increase in LAMs was found in obese SAT than obese VAT, although no significant difference was found between lean and obese patients.⁵⁰ Similar to observations in obese mice, human LAM subsets express the signature TREM2, CD9, and lipoprotein lipase (LPL) genes, exhibit a lipid-laden phenotype, and contain lipid-lysosomal structures. These subsets are localized to CLS, which positively correlate with a high body mass index (BMI), in human AT, displaying profiles of phagocytosis, endocytosis, and lipid metabolism.⁴²

The IM subsets were found to be increased in obese human SAT and characterized by signature expressions of CCL3L1, TNF, and C-X-C motif chemokine ligand 3 (CXCL3). In contrast to the LAM subsets, the IM subsets are suggested to originate from Mo-1 subsets under the influence of pro-inflammatory cytokines and JUN-FOS signaling pathways, with Mo-2 subsets acting as an intermediate phase. Both the IM and LAM subsets are implicated in promoting AT inflammation through the secretion of TNF, IL-1β, IL-18, CXCL8, platelet-derived growth factor subunit B (PDGFβ), and TNF superfamily member 13b (TNFSF13B), which can elicit responses from stromal cells, DCs, and circulating myeloid subsets.⁴⁷ Consistent with those findings, a spatial transcriptomic analysis indicated that pro-inflammatory human ATMs tend to cluster near DCs and monocytes, whereas anti-inflammatory human ATMs are more dispersed throughout the tissue.⁴⁹

A recent integrated analysis combining single-cell sequencing, single-nucleus sequencing, and spatial transcriptomics across various human AT depots (subcutaneous, omental, and perivascular) has identified additional subsets. For instance, redox-regulatory metabolic macrophages (Mox) were uniquely found in human SAT and characterized by the signature genes heme oxygenase 1 (HMOX1), IL1B, and NRF2.³⁸ Additionally, MMe subsets were enriched in SAT, with smaller proportions detected in omental and perivascular AT. LAM subsets were observed in both SAT and omental AT, and several clusters of M2-like cells were enriched in omental AT. This comprehensive analysis underscores the heterogeneity of human ATM subsets across different fat depots.

Single-cell studies highlight the ongoing challenges of understanding the in vivo functionality of ATM subsets in obesity. A previous review discussed the potential functions and crosstalk of these subsets.⁵¹ ATMs are involved in numerous functions crucial for AT remodeling in obesity. Despite the efforts of researchers, determining the precise regulatory roles and overall functions of ATMs remains difficult. Here, we provide a summary of ATM functions and mechanisms in context-specific situations during obesity (Fig. 2).

Production and secretion of inflammatory factors

ATMs are known to produce and secrete both anti- and pro-inflammatory factors, including cytokines and chemokines.^5,21,27 In steady states, resident ATMs generate anti-inflammatory cytokines such as IL-10 and IL-4, which are crucial for maintaining AT insulin sensitivity and overall homeostasis.^52,53 In obesity, ATMs shift to a pro-inflammatory state that is closely linked to insulin resistance, although the precise mechanisms remain unclear. Obese ATMs secrete pro-inflammatory cytokines such as TNF-α, IL-6, and IL-1β, which might drive adipocyte dysfunction and promote AT inflammation.^3-5,10,54,55 Furthermore, obese ATMs and adipocytes secrete inflammatory chemokines such as monocyte chemoattractant protein-1 (MCP-1)/CCL2 and macrophage inflammatory protein 1α (MIP-1α)/CCL3, which contribute to the recruitment of monocyte-derived macrophages, potentially leading to the formation of CLS.^55,56 Factors such as adipocyte hypertrophy, local hypoxia, and adipocyte-derived inflammatory factors—including chemokines such as regulated on activation, normal T-cell expressed and secreted (RANTES)/CCL5, MCP-2, MCP-3, and CXCL14—have also been proposed as triggers for pro-inflammatory ATM accumulation.^4,8,56,57 Furthermore, increased expression of TLRs and inflammasomes, such as TLR4 and nod-like receptor protein 3, has been observed in obese ATMs in response to elevated levels of damage-associated molecular patterns and pathogen-associated molecular patterns such as oxidized low-density lipoprotein and LPS. Such activation promotes signaling through nuclear factor κB (NF-κB), signal transducer and activator of transcription 1 (STAT1), and caspase-1, thereby inducing the production of pro-inflammatory cytokines and chemokines.^54,58,59

Interestingly, calorie restriction or weight loss improves the metabolic parameters associated with obesity, but the ATM profile does not fully reflect those changes. A history of obesity independently exacerbates metabolic profiles during weight regain, implicating ATMs as key mediators.^43,60-62 The concept of trained innate immunity suggests that epigenetic reprogramming in obese ATMs persists during weight loss, potentially leading to enhanced inflammatory responses upon subsequent obesity challenges.^63-65 Studies in mice fed a Western diet have linked epigenetic reprogramming of macrophage progenitors in BM to persistent inflammation.⁶⁶ Thus, ATMs and BM progenitors in formerly obese individuals might have imprinted immunological memory that is associated with higher inflammatory responsiveness.^61-63

Adaptive immune response

Increased pro-inflammatory ATMs are implicated in activating the adaptive immune response, potentially creating a feedback loop in which CD4⁺ and CD8⁺ T-cell numbers rise in obese AT and promote ATM accumulation.^67,68 ATMs possess antigen-presenting capabilities that facilitate T-cell clonal expansion within AT.^69,70 Through phagocytosis, ATMs acquire local antigens and present them to naïve CD4⁺ T cells, promoting the proliferation of antigen-specific CD4⁺ T cells within AT.^70,71 Obese ATMs have been demonstrated to skew naïve CD4⁺ T-cell polarization toward pro-inflammatory T helper (Th)-1 and Th17 CD4⁺ T-cell subsets, aligning with the obese AT T-cell (ATT) profile.⁶⁹ Meanwhile, the accumulation of CD8⁺ T cells has been suggested to precede the increase in ATMs and coincide with elevated levels of pro-inflammatory cytokines, such as interferon-γ, that act as signals for the recruitment and activation of ATMs.⁶⁸ Both resident and pro-inflammatory ATMs have been implicated as significant contributors to the generation of adaptive immune memory.⁷¹ Therefore, changes in ATM diversity during obesity may modulate the ATT profile, influencing the progression of AT inflammation.

Lipid-buffering

In steady states, ATMs maintain metabolic homeostasis by acting as lipid buffers that absorb extracellular lipids, including FFAs secreted by adipocytes and lipid remnants from dead adipocytes.^33,55,72,73 This role is supported by evidence demonstrating that PDGFcc, a receptor involved in lipid storage, is expressed on TIM4⁺ resident macrophages and regulates energy storage in AT.⁷⁴ In enlarged adipocytes, ATMs enhance lipid storage capacity by increasing the uptake of excess TG and FFA, thereby aiding in preventing ectopic lipid accumulation and its associated lipotoxicity effects in other tissues.^73,75,76 This role is further exemplified by lipid-laden ATMs, identified as MMe or LAM subsets, that are characterized by intracellular lipid droplets and induce lysosomal biogenesis.^36,76-78 These lipid-laden ATMs surround dead adipocytes, forming CLS that facilitate the clearance of dead fat cells.^9,75,79 The internalization of large amounts of lipids or cell debris by ATMs increases lipid catabolism by inducing lysosomal activity via vesicle fusion with the primary and secondary lysosomes.^20,78 Though the fate of lipids post-fusion remains unclear, this process could activate pro-inflammatory responses and shift the energy metabolism of ATMs.⁷⁹ Notably, activated lysosomal biogenesis correlates positively with insulin resistance progression and ATM accumulation.⁷⁸

Mitochondrial uptake

Recent studies indicate that in addition to phagocytosing dead cells and debris, ATMs control AT metabolic homeostasis by uptaking mitochondria released by neighboring adipocytes.⁸⁰ This phenomenon occurs in both white AT and brown AT, where ATMs remove damaged mitochondria contained within extracellular vesicles (EVs).^81,82 Dietary factors appear to influence this regulation: a lard-based HFD rich in long-chain fatty acids inhibits ATMs’ mitochondrial acquisition, increasing mitochondrial transfer from adipocytes to circulation and other organs.⁸³

Mitochondrial uptake by ATMs in white AT has been suggested to depend on heparan sulfates, which selectively bind mitochondria.⁸² Obese ATMs exhibit reduced heparan sulfate expression and lower mitochondrial transfer rates, linking heparan sulfate expression to the mitochondrial transfer process. Furthermore, deletion of myeloid-Ext1, the gene responsible for heparan sulfate biosynthesis, reduces mitochondrial uptake by ATMs and worsens obesity by increasing adiposity and insulin resistance and decreasing energy expenditure.⁸²

Mitochondrial uptake leads to a transcriptionally distinct ATM population characterized by upregulated genes related to mitochondrial-DNA, chemokines, and anti-inflammatory responses and downregulated MHC-II antigen-presenting genes.⁸² ATMs acquiring mitochondria show gene enrichment in the hypoxia-inducible factor 1 (HIF-1α) pathway and reduced expression in the electron transport chain and collagen synthesis pathway. Nonetheless, the specific role of this ATM subset in altering ATM diversity and function in obesity post-mitochondrial uptake requires further exploration. Overall, these insights underscore the role of ATMs in regulating AT homeostasis via mitochondrial uptake.

Exophagy

The enhanced ability of obese ATMs to internalize lipids and cell debris is accompanied by an ability to secrete EVs, including those from lysosomal exocytosis.^6,78 ATMs at CLS develop large lysosomal compartments with low pH and active lysosomal hydrolase enzymes, which facilitates extracellular catabolism and the uptake of particles from dying adipocytes.^6,75,84 Thus, ATMs at CLS could be activated by materials from dying adipocytes, leading to pro-inflammatory cytokine release by ATMs⁸⁵ that results in a feed-forward mechanism promoting AT inflammation.

Obese ATMs secrete EVs containing miRNA that can be internalized by adipocytes, affecting insulin signaling and metabolism.⁸⁵ Treating insulin-sensitive mice with obese ATM-derived EVs induces systemic insulin resistance and glucose intolerance, whereas treating obese insulin-resistant mice with lean ATM-derived EVs improves glucose tolerance and insulin sensitivity.^86,87 Specific miRNAs have been found to be enriched in EVs from different types of ATMs, such as miR-690 from M2-type ATMs and miR-155 from M1-type ATMs.^85,87 Although pinpointing exact cell sources is challenging, exosome signaling in AT extends beyond local regulation. Uptake of AT-derived exosomes by monocytes promotes their pro-inflammatory activation, inducing insulin resistance.⁸⁸ Notably, the miR-34a in obese AT-derived EVs regulates ATM polarization by suppressing M2-type polarization and promoting AT inflammation and insulin resistance progression.⁸⁶

Angiogenesis and adipogenesis

ATMs are crucial in AT remodeling, regulating both angiogenesis and adipogenesis.^89,90 Angiogenesis has been suggested to prevent hypoxia during AT expansion. Macrophage depletion via clodronate liposomes reduces blood vessel formation in AT, indicating ATMs’ role in promoting angiogenesis.⁹¹ Specifically, LYVE-1⁺ ATMs in the tip regions of adult gonadal AT are recruited and activate angiogenesis through the vascular endothelial growth factor (VEGF)-VEGFR2 pathway and the secretion of matrix metalloproteinases (MMPs) such as MMP-7, MMP-9, and MMP-12.⁹² Another study demonstrated that human ATMs secrete MMPs and promote endothelial cell tube formation in Matrigel systems, indicating their angiogenic role.⁹³ Increased ATM infiltration is associated with higher levels of angiogenic factors such as TNF-α and PDGF.^89,94 Local hypoxia has been implicated as an activator of obese ATMs, potentially altering their functionality.⁸⁹ Nonetheless, the precise regulation of new vessel formation distribution remains poorly understood.

ATMs are also known to participate in adipocyte turnover by initiating adipogenesis. In lean AT, M2-type ATMs expressing osteopontin promote tissue repair and form an adipogenic niche by recruiting platelet-derived growth factor receptor α (PDGFRα)⁺ adipocyte progenitor cells to sites of dying adipocytes.⁷² Osteopontin-deficient mice, which cannot form regenerative adipogenic regions, showed reduced pre-adipocyte differentiation and increased pro-inflammatory ATMs.⁷² The release of the pro-inflammatory cytokines TNF-α and IL1-β by M1-type ATMs could inhibit adipogenesis by suppressing pre-adipocyte differentiation,^95-97 which leads to adipocyte hypertrophy and eventually cell death.⁹

Fibrosis

In obese AT, extracellular matrix (ECM) remodeling, including excessive production and degradation of ECM proteins, leads to irreversible fibrosis.⁹⁸ Fibrosis exacerbates tissue dysfunction in obese, insulin-resistant AT, with ATMs implicated as crucial modulators of fibrosis development. They contribute significantly by secreting the pro-fibrotic factors transforming growth factor β1 (TGFβ1) and PDGF, which activates myofibroblast-like cells from AT progenitors.^99,100 Furthermore, pro-inflammatory activation of ATMs is closely associated with modulating fibrosis. In vitro studies demonstrate that LPS-activated macrophages increase ECM production in human preadipocytes.⁹⁷ Meanwhile, pro-inflammatory cytokines (TNF-α and IL-1β) from macrophages induce MMP expression in both preadipocytes and adipocytes, contributing to ECM breakdown.^101,102

Activation of TLR4 signaling in ATMs, particularly at CLS, upregulates fibrosis-related genes via macrophage-inducible C-type lectin induction.^103,104 Obese TLR4-deficient mice showed reduced collagen deposition and downregulated collagen-related genes, highlighting TLR4’s role in promoting fibrosis.¹⁰⁵ Furthermore, pro-inflammatory ATMs are implicated in hypoxia-induced HIF-1α accumulation through their increase of nitric oxide production.¹⁰⁶ HIF-1α activation triggers profibrogenic gene transcription in preadipocytes, worsening ECM production and AT fibrosis progression.^106,107 Conversely in human AT, most ATMs, not just those located at CLS, contribute to fibrosis. M2-type ATMs dispersed in fibrotic areas expressed high TGFβ levels, and co-culture of ATMs and adipocytes increased TGFβ expression and activity in both cell types, suggesting a communication loop that promotes fibrosis development in AT.¹⁰⁸

Given their critical roles in controlling AT function and metabolism during obesity, ATMs are key targets for preventing and treating obesity-related metabolic diseases. Current strategies focus on manipulating ATM functionality, metabolism, and molecular regulation to reverse their pro-inflammatory responses. Inhibiting ATM activation, including recruitment, polarization, and proliferation, could prevent AT inflammation and mitigate the harmful effects of obesity (Fig. 3). Previous studies have demonstrated that treating adipocytes with IL-10 suppresses TNF-α-induced insulin resistance, and administering IL-4 to obese mice reduces AT inflammation and enhances insulin sensitivity.^8,27,109

Investigating the regulatory mechanisms of the AT niche in shaping ATM phenotypic adaptability and function remains warranted. Obese ATMs exhibit unique activation in response to increased levels of full-length oxidized phospholipids, inducing pro-inflammatory gene expression and bioenergetic activation.¹¹⁰ Targeting the regulation of antioxidant enzymes in AT could modulate ATM activation. Moreover, targeting ATM mitochondrial function has shown promise in regulating energy metabolism and alleviating obesity-induced AT inflammation.^{82,87,111,113} Additionally, manipulating ATM responses to metabolic hormones such as insulin influences their activation and cytokine production.^{52,53,114,115}

Pro-inflammatory TREM2-expressing ATM subsets (LAMs) in obesity are a potential therapeutic target. However, studies involving BM transplantation indicate that TREM2 expression might serve merely as a marker; it is dispensable for regulating metabolic profiles in obesity.¹¹⁶ The regulation of lipid uptake and inflammatory pathways in ATMs remains poorly understood. Silencing lipid uptake proteins such as LPL and CD36 in obese ATMs leads to ectopic lipid accumulation, whereas promoting lipid storage in macrophages is associated with reduced pro-inflammatory responses.^{76,113,117-119} Conflicting studies report impaired lipid droplets in macrophages without a change in AT inflammation.¹²⁰ Despite those discrepancies, the similar expression of TNF-α, IL-1β, ABCA1, PLIN2, and CD36 in obese human and mouse ATMs highlights the need for further research to clarify their regulatory mechanisms in obesity.^20,36,47

The reduced cell egress of ATMs in obesity highlights their potential to have long-term regulatory effects.²⁹ ATMs derived from weight loss exhibit primed bioenergetic metabolism and increased production of pro-inflammatory cytokines.⁶² Altered epigenetic mechanisms in ATMs and their progenitors during obesity are thus implicated in prolonged heightened responses that exacerbate metabolic dysfunction.^63,65,121 These observations point to the potential for precision medicine targeting epigenetic regulation in ATMs to prevent and treat obesity-associated metabolic diseases.

Research on ATMs has illuminated the significance of immunometabolism in regulating obesity-associated metabolic diseases. Immersive characterization of ATM subpopulations at the single-cell level emphasizes the need to clarify their regulatory mechanisms during obesity and their effects on overall AT function, particularly in humans. Notably, determining how energy imbalance affects the AT microenvironment and triggers ATM activation is crucial. Future work is needed to comprehend and translate findings on ATM biology and functionality from mouse models to human physiology. Better understanding of existing treatments for obesity-induced insulin resistance, such as thiazolidinediones and incretin-based hormones, will require the identification of the triggers for ATM activation and inflammatory profiles. Therefore, elucidating the orchestration of ATM phenotypes and functionality remains a priority for future research.

The authors declare no conflict of interest.

This work was supported by the National Research Foundation of Korea (NRF-2022R1A2C2006224) and MunSeok Research Grant (Grant No. KSSO202001) from Korean Society for the Study of Obesity. All figures in this article were created using BioRender.

Study concept and design: SS and KWC; acquisition of data: SS, VJ, and HN; analysis and interpretation of data: SS, VJ, and HN; drafting of the manuscript: SS and VJ; critical revision of the manuscript: SS, VJ, HN, and KWC; obtained funding: KWC; and study supervision: KWC.