Adipose tissue inflammation and metabolic dysfunction in obesity

Abstract

Several lines of preclinical and clinical research have confirmed that chronic low-grade inflammation of adipose tissue is mechanistically linked to metabolic disease and organ tissue complications in the overweight and obese organism. Despite this widely confirmed paradigm, numerous open questions and knowledge gaps remain to be investigated. This is mainly due to the intricately intertwined cross-talk of various pro- and anti-inflammatory signaling cascades involved in the immune response of expanding adipose depots, particularly the visceral adipose tissue. Adipose tissue inflammation is initiated and sustained over time by dysfunctional adipocytes that secrete inflammatory adipokines and by infiltration of bone marrow-derived immune cells that signal via production of cytokines and chemokines. Despite its low-grade nature, adipose tissue inflammation negatively impacts remote organ function, a phenomenon that is considered causative of the complications of obesity. The aim of this review is to broadly present an overview of adipose tissue inflammation by highlighting the most recent reports in the scientific literature and summarizing our overall understanding of the field. We also discuss key endogenous anti-inflammatory mediators and analyze their mechanistic role(s) in the pathogenesis and treatment of adipose tissue inflammation. In doing so, we hope to stimulate studies to uncover novel physiological, cellular, and molecular targets for the treatment of obesity.

INTRODUCTION

Obesity is now considered as a worldwide epidemic (1) and a strong risk factor for insulin resistance, type 2 diabetes mellitus (T2DM), cardiovascular diseases (CVD), immune disorders, and nonalcoholic fatty liver disease (NAFLD), in addition to several type of cancers. Overall, obesity is associated with a reduction of quality of life, shortened life span, and increased healthcare costs (2–5). Because of a pathophysiology that recapitulates the more complicated picture of multifactorial chronic disease similar to the aging process, obviously obesity cannot be simply considered the result of an energy imbalance between calorie intake and expenditure. Thus, a host of metabolic abnormalities, oxidative stress, mitochondrial dysfunction, immune dysfunction, and chronic low-grade inflammation, have been identified in the overweight obese organism (6, 7). Although it has been almost universally established that adipose tissue (AT) responds to overnutrition by mounting an immune response, the initial inflammatory trigger remains unfortunately unknown. As a result, the clinical efficacy of drugs targeting the presently discovered inflammatory pathways has been disappointing. Perhaps, as discussed later in this review, little attention has been devoted to endogenous anti-inflammatory mediators, other potentially intriguing therapeutic targets in AT inflammation. In this review, we summarize the most recent findings about AT inflammation and obesity, highlighting physiological, cellular, and molecular mechanisms through which AT inflammation contributes to AT dysfunction and related systemic complications.

THE INFLAMMATORY PHENOTYPE OF WHITE ADIPOSE TISSUE

White adipose tissue (WAT) is the major fat-storing depot and also serves as the largest endocrine organ to secrete adipokines and cytokines systemically. Adipokines are involved in various metabolic and physiological signaling cascades and regulate insulin signaling, glucose uptake, fatty acid oxidation, and other energy-producing and metabolic processes (8). Cytokines regulate inflammation and resolution of inflammation along with adaptive and reparative angiogenesis. Weight gain and obesity cause a phenotypic switch of WAT, which is characterized by the appearance of inflamed, dysfunctional adipocytes along with infiltration of immune cells in the stromal vascular fraction (9, 10). Inflamed adipocytes secrete, both locally and systemically, proinflammatory cytokines, which in turn disrupt the normal function of AT itself as well as that of remote organs (11). From this standpoint, AT can be considered an immune and secretory organ, and obesity an inflammatory immune disease (Fig. 1).

Figure 1.

The inflammatory phenotype of expanding adipose tissue. Hypertrophic adipocytes and tissue resident immune cells experience phenotype changes that halt secretion of anti-inflammatory, protective cytokines to begin secretion of inflammatory adipokines and cytokines that act both locally and systemically to induce peripheral insulin resistance. The inflammatory reaction is sustained by adipocyte-derived chemoattractants such as C-C chemokine receptor type 2 (CCR2), monocyte chemoattractant protein (MCP), and semaphorin 3A (SEMA3A).

Both animal and human studies have confirmed the association between increased adiposity and AT inflammation following excessive caloric intake. Using several immunocompromised mouse models, Lee et al. (12) demonstrated the essential role of inflammation in the establishment of insulin resistance following long-term consumption of an obesogenic Western-type diet. A unique feature of the inflammatory responses of expanding WAT is its duration and intensity, i.e., a persistent, low-grade inflammation that fails to resolve. Such obesity-related chronic low-grade inflammation and subsequent altered metabolism have been termed “metaflammation” (9). Inflammation in general is an energy-wasting process that enhances energy expenditure and reduces energy intake in direct and indirect manners. Directly, inflammatory cytokines, such as TNFα, IL-1, and IL-6, induce energy expenditure by binding to signaling receptors located in the central nervous system or in the tissue of metabolic active organs. Indeed, they are provided with leptin-like properties that promote energy expenditure (13–15). Induction of leptin expression represents a molecular mechanism of the inflammatory activity. Leptin expression is increased in adipose tissue by inflammation. Leptin transcription is induced by hypoxia (16) and inflammatory mediators (17), all of which are experienced by expanding adipose depots. Furthermore, leptin receptor expression is induced by TNFα (18), which provides a mechanism by which proinflammatory cytokines enhance leptin activity for energy expenditure. Leptin is an adipokine that inhibits appetite and induces energy expenditure (19), thus indirectly increasing energy expenditure.

Surprisingly, the unique features of AT inflammation induced by overnutrition are not associated with a significant increase in energy expenditure, which permits the coexistence of inflammation and weight gain in obese people (20). Nonetheless, inflammation of AT still shares several commonalities with the traditional inflammatory response, that is, the infiltration of bone marrow-derived immune cells and the secretion of inflammatory mediators including, but not limited to, chemokines and cytokines by adipocytes and resident immune cells. Moreover, because of its conspicuous mass and distribution, even in the apparently healthy normal-weight individual, the inflamed WAT can cause widespread systemic inflammation via release of cytokines (21).

Based on its anatomic location and morphological structure (reviewed in Ref. 22), WAT develops diverse and unique inflammatory phenotypes. Accordingly, it is currently well established that obesity instigates a more complex and intense inflammatory reaction in visceral WAT (VAT) than in subcutaneous WAT (SAT). Indeed, VAT contains more macrophages compared with SAT in obese mice (23–25) and obese humans (26–28). With obesity and insulin resistance, VAT adipocytes experience a much higher degree of hypertrophy than SAT both in humans (29, 30) and in animal models (31). Furthermore, VAT inflammation in obese humans is associated with decreased expression of lipogenic markers (32), probably due to the fact that more cells are switched to an inflammatory rather than a lipid storage phenotype, which leads to the development of metabolic complications, such as ectopic lipid deposition in skeletal muscle and liver (32). Since ectopic lipid deposition dampens peripheral insulin signaling, VAT inflammation is considered to have a major impact on obesity-related metabolic disorders such as systemic insulin resistance and development of T2DM (29, 33, 34). While these lines of research support a higher involvement of VAT inflammation in obesity-related complications, it should be nonetheless noted that many investigators have reported convincing evidence of a role for SAT inflammation in metabolic complications (34–36). The ongoing controversy on the role that different WAT depots play in metaflammation has been partially addressed in mice. To explain the mechanism underlying the different impacts of VAT and SAT on metabolism, Rytka et al. (37) artificially increased fat mass in mice by transplanting epididymal VAT obtained from littermate C57Bl6/J donor mice into the parietal peritoneum that drains in the caval system or the mesenterium that drains in the liver portal system. The procedure induced AT inflammation in both experimental groups of transplanted mice. However, only the mice with mesenterium-transplanted fat experienced elevated IL-6 levels and free fatty acid (FFA) in the portal vein and developed impaired glucose tolerance (37). Human studies of meal FFA uptake further confirm the heterogeneity in the metabolism of VAT and SAT. Expressed relative to the same mass of AT, meal FFA uptake is greater in intra-abdominal than in abdominal subcutaneous fat in both sexes (38, 39). The direct uptake of plasma FFA is also greater in omental than in abdominal subcutaneous fat of women (40). These data suggest that the anatomic location and the unique structural features of VAT enable it to directly feed inflammatory cytokines and metabolites to important organs, such as the liver, that play a role in the regulation of insulin action and systemic metabolism.

In the next sections, we will summarize our current understanding of the cellular and molecular mechanisms implicated in the inflammatory response of the WAT in obesity.

THE ROLE OF BONE MARROW ADIPOSE TISSUE

In healthy adults, the approximatively 10% of adipose tissue located in the bone marrow also regulates whole body energy metabolism, along with local and systemic inflammatory responses. The role of bone marrow adipose tissue (MAT) in energy metabolism and inflammation has been recently reviewed (41, 42); thus, here we will briefly discuss key relevant aspects of MAT’s role in obesity. In terms of metabolic responses, MAT expresses and secretes both adiponectin and leptin. Whereas in vitro studies have reported that primary cultured MAT express a lower level of adiponectin (43, 44), studies in mice and humans have found higher levels of adiponectin expression and secretion in MAT over WAT (45). Furthermore, modulation of MAT mass is correlated to serum adiponectin abundance following caloric restriction (45) or treatment with thiazolidinediones (46). MAT also expresses and secretes leptin (47, 48).

The contribution, though, of bone marrow adipocytes to the metabolic and inflammatory complications of obesity remains somewhat controversial. Differently than visceral fat adipocytes, bone marrow adipocytes are characterized by lower expression levels of adipose-specific genes such as peroxisome proliferator-activated receptor (PPARγ) and fatty acid-binding protein-4 (FABP4), and higher abundance of inflammatory response signaling pathways. In vitro studies have shown that primary cultured bone marrow adipocytes secrete significant levels of IL-6 but only small levels of IL-1 and TNFα (49). Others instead have reported increased expression of inflammatory response genes such as TNFα, IL-6, and IL-1β (43). With high-fat feeding, mice experience increases bone marrow adiposity with adipocytes that surprisingly express decreased inflammatory genes such as TNFα and IL-1β, (50). It should be noted, though, that MAT is unquestionably the prevalent source of immune cells for other adipose depots and metabolic organs in obesity. Thus, it is reasonable to conclude that cytokine production by MAT adipocytes might be mainly important to regulate bone marrow kinetics and that MAT’s contribution to systemic inflammation in obesity is indirectly supported by production and release of immune cells.

MACROPHAGES, MASTER REGULATORS OF WAT INFLAMMATION

Infiltration of bone marrow-derived inflammatory cells is a key feature of the WAT metaflammation. It is well established that infiltration of inflammatory cells in expanding WAT almost invariably causes adipocyte dysfunction and metabolic dysfunction, such as glucose intolerance and insulin resistance. AT macrophages (ATMs) were the first immune cell population to be discovered and studied in this process (51), although more recent reports have emphasized the role of earlier neutrophil infiltration (discussed later). During local and systemic inflammatory responses, tissue resident macrophages present antigens to initiate recruitment of other immune cells and secrete cytokines to regulate inflammatory signaling cascades in the host tissue. This physiological role of macrophages appears to be maintained also in the metaflammation of expanding WAT depots (51, 52). The relevance of this phenomenon is confirmed by several clinical reports demonstrating ATM infiltration in the WAT of obese humans (28, 34, 53–56). Furthermore, clinical studies have also clarified the relationship between ATM infiltration in WAT and insulin resistance. WAT obtained from insulin-resistant and obese patients contains more macrophages with increased expression of proinflammatory mediators (29), and this phenomenon strongly correlates with metabolic dysfunction (33, 57).

Activated ATMs are the main source of proinflammatory mediators such as TNFα, inducible NO synthase (iNOS, monocyte chemoattractant protein-1 (MCP-1), and IL-6 (51, 58). Hotamisligil et al. (59) initially reported in 1993 that the gene and protein expression of TNFα are upregulated in AT, especially in ATMs, obtained from obese animal models. This first observation was followed by many confirmatory studies in laboratory animals and humans showing the upregulation of proinflammatory adipokines including TNFα, IL6, IL-18, and MCP-1 in obesity and their role in the development of insulin resistance and T2DM (51, 58–61). The importance of cytokines secreted by ATMs in whole body metabolism was mechanistically elucidated by Aouadi et al. (62), who showed that ATM-specific silencing of TNFα and osteopontin using siRNA improves insulin sensitivity and glucose tolerance. Osteopontin is a more recently studied inflammatory mediator that appears to favor infiltration, survival, and proliferation of monocytes/macrophages in WAT (63). IL-1β produced by the NOD-, LRR-, and pyrin domain-containing 3 (NLRP3) inflammasome regulates adipose tissue inflammation during obesity (64–66). And inhibition of NLRP3 reduces proinflammatory cytokines and subsequent macrophage infiltration (67). Proteoglycans are also known to regulate inflammation. A recent report revealed that, in obese condition, adipocytes produce versican and macrophages produce biglycan, and the cross-talk between these proteoglycans affects adipose tissue inflammation and insulin resistance (68).

Recent studies have also emphasized the existence of a complex interplay between WAT signaling and the central nervous system (CNS), an organ that contributes to regulation of metabolism at multiple steps (reviewed in Ref. 69). A recent line of research demonstrates that factors involved in axonal growth are also implicated in WAT inflammation. Neuroimmune guidance cue netrin-1 and semaphorin 3E (SEMA3E) with its receptor olexinD1 are regulators of the development of the neuronal system that are upregulated in the obese WAT, where they function as chemoattractants for macrophages (70–72), also by inhibiting macrophage egression (73). Similarly, the neuronal cytokine fractalkine (CX3CL1) and its receptor modulate monocyte adhesion to adipocytes (74).

The classic accepted paradigm is that tissue resident macrophages originate from bone marrow monocytes that infiltrate tissue during physiological immunosurveillance or in response to inflammatory events. Interestingly, diet-induced obesity increases the circulating levels of CD11b(+) monocytes, which express the chemoattractant receptor leukotriene B (75) receptor (BLT-1), and BLT-1 has been shown to sustain monocyte trafficking to WAT (76). More recently, it has been shown that proinflammatory ATMs express genes involved in myelopoiesis and immune cell recruitment, a process that affects circulating levels of monocytes and neutrophils (77, 78). Thus, obesity generates a self-feeding cycle of monocyte/macrophage infiltration to sustain low-grade chronic inflammation of WAT.

While this classic monocyte/macrophage recruitment remains a confirmed mechanism of inflammatory cell infiltration in expanding WAT, a new emerging concept in the field is that of local macrophage proliferation. Cytokines such as IL-4 are reported to stimulate local macrophage proliferation in AT (79). This local proliferation of ATMs occurs mainly at crown-like structures (CLS) that surround necrotic adipocytes, resulting in a preferential increase of M2 macrophage in WAT (80). The consequence and impact of local ATM proliferation, especially of the M2 type, in the pathophysiology of AT dysfunction and metaflammation remains uncharacterized and fertile ground for future research in the field.

The phenotype of obese ATMs is polarized to a proinflammatory state, a phenomenon that has been associated with metabolic complications (57, 81–83). In lean, insulin-sensitive mice, ATMs express anti-inflammatory genes such as IL-10 and arginase-1, whereas ATMs in obese mice highly express proinflammatory genes such as TNFα and iNOS (84). Similarly, the major characteristic of WAT obtained from patients with metabolic abnormalities, such as T2DM patients, is adipocyte hypertrophy with high infiltration of proinflammatory macrophages (35). These “metabolically activated” proinflammatory ATMs function within the context of low-grade inflammation and therefore appear to have important distinctions compared with classically activated M1 macrophages that occur within the context of a full-fledged inflammatory reaction. In fact, whereas M1 macrophages in acute inflammatory reactions highly express CD38, CD274, and CD319 on their cell surface, the ATMs of the obese organism do not (85); they actually show a less inflammatory phenotype with a different gene expression profile compared with classically activated M1 macrophages (86). As result, the expression of proinflammatory cytokines such as TNFα and IL-6 in obese ATMs remains lower than that of classically activated M1 macrophages (85). Although these observations are consistent with the occurrence of low-grade inflammation in obesity, the question remains as to how this lower degree of immune cell activation is achieved and maintained. Current research supports the concept that the concomitant activation of endogenous anti-inflammatory cascades is responsible for dampening macrophage activation and overall WAT inflammation in obesity. Thus, recruitment of anti-inflammatory macrophages via increased IL-10 expression has been observed in WAT after a high-fat diet (HFD) (81), and deletion of IL-10 from the hematopoietic system worsened HFD-induced inflammation and insulin resistance, as one might have predicted (87). Similarly, we have observed increased expression levels of the proangiogenic and anti-inflammatory cytokine IL-19 in stromal vascular fraction and visceral adipocytes of mice given a HFD (our unpublished observation and Fig. 2). Others have implicated a role for IFNγ, another key regulator of macrophage activation and phenotypic switches. Thus, IFNγ knockout mice exhibit less inflammatory ATM in conjunction with decreased adipocyte size and increased insulin sensitivity following HFD (88). Furthermore, interferon-regulatory factor (IRF)-5 expression is upregulated in obesity, and IRF-5 expression is positively correlated with macrophage infiltration and the secretion of proinflammatory adipokines (89).

Figure 2.

With increased energy storage and free fatty acid (FFA) cargo, white adipocytes undergo abnormal expansion, which results in hypoxia and remodeling-induced senescence. Hypoxia and senescence initiate and sustain chronic low-grade inflammation. Under these conditions, adipocytes experience endoplasmic reticulum (ER) stress and increased reactive oxygen species (ROS) production. Dysfunctional adipocytes also secrete inflammatory cytokines at the expense of production of protective adipokines, such as adiponectin. Noteworthy, inflamed adipose depots also express anti-inflammatory mediators, such as IL-10 and IL-19, whose role in the overall regulation of the immunometabolic response of adipose tissue remains largely unexplored.

Molecular mechanisms operating within macrophages themselves have also been investigated. Thus, it has been reported that suppression of macrophage activation in obesity is partially controlled via adenosine receptors expressed on macrophages and that these receptors could respond to accumulation of local catabolites (90, 91). In a similar line of research, the inositol-requiring enzyme 1α (IRE1α), a key regulator of endoplasmic reticulum (ER) stress in metabolic organs, has been recently investigated. Shan et al. (92) have implicated IRE1α in the regulation of WAT inflammation by showing that myeloid-specific IRE1α knockout mice have a reduced inflammatory phenotype and are protected from HFD-induced obesity, insulin resistance, hyperlipidemia, and hepatic steatosis (92). More clinically relevant, PPARγ has also been implicated in regulation of insulin resistance as well as inflammation. PPARγ is essential for modulation of metabolically activated macrophages, suggesting that PPARγ may increase macrophages with an anti-inflammatory phenotype, and therefore, it improves glucose metabolism via an anti-inflammatory mechanism in addition to its direct metabolic actions (93, 94).

Despite this wealth of knowledge, currently anti-inflammatory therapies have unfortunately failed to correct the metaflammation of obesity, which poses both a puzzling research question and a therapeutic obstacle. To help explain these unexpected negative results, Scherer and colleagues (95) have recently demonstrated that inflammatory cytokines signaling is paradoxically required for appropriate expansion and metabolic flexibility of WAT. They found that congenital silencing of TNFα or IL-1β function in diet induced obese mice further exacerbates insulin resistance and glucose intolerance due to loss of compensatory angiogenesis and insufficient WAT remodeling. Consistent with this study, clinical data have yielded contrasting evidence on the effect of TNFα and IL-1β blocking therapy on glycemic control and insulin sensitivity in obese, diabetic patients. Thus, systemic blockade of TNFα does not improve insulin resistance in humans, and blockade of IL-1β with canakinumab was not associated with any reduction in the rate of incident diabetes in prediabetic insulin resistance patients (96, 97). Overall, how the balance between pro- and anti-inflammatory cytokines ultimately determines whether or not an expanding adipose depot develops a dysfunctional degree of inflammation which impairs its metabolic activity remains insufficiently explored and poorly understood. Further studies are needed to fully understand the link between cytokines and metabolism to identify new therapeutic targets.

INFILTRATION OF OTHER IMMUNE CELLS

Whereas macrophages are the most studied and undoubtedly important immune cell type involved in AT inflammation, other immune cells such as neutrophils and lymphocytes appear to play relevant mechanistic roles (98, 99) (Fig. 1). Animal studies demonstrate that administration of a HFD to lean mice causes a very early infiltration of neutrophils in VAT before the onset of insulin resistance and obesity (100). These infiltrated neutrophils initiate recruitment of macrophages and other antigen-presenting cells. Bidirectional cross-talk between adipocytes and neutrophils has been recently reported to cause WAT inflammation via IL-1β production with the involvement of infiltrating macrophages (101). This neutrophil-dominated early inflammatory cascade was found causal to the development of hepatic insulin resistance and metabolic disorders (102, 103). Prevention of WAT neutrophil infiltration by inhibiting cytosolic phospholipase-A_2α (cPLA2a) or the endothelial cell adhesion molecule ICAM-1 reduced TNFα secretion from adipocytes, preserved hepatic insulin signaling, and prevented abnormal gluconeogenesis (102). Neutrophil elastase secreted from infiltrated neutrophils was found to be one of the key components affecting metabolic functions. Neutrophil elastase inhibitors or the deletion of neutrophil elastase in mice preserved insulin sensitivity and attenuated ATMs infiltration in HFD-fed mice (103).

The infiltration of T cells into expanding adipose depots is also considered to be another important, although controversial (12), mechanism for macrophage recruitment and WAT inflammation (104, 105). Similar to ATMs, views on the immunomodulation of T cells in expanding adipose depots remains somewhat controversial, especially for the role of regulatory T cells (Tregs). Some studies reported downregulation of Tregs in obese states, potentially resulting in chronic WAT inflammation and subsequent insulin resistance (106–109). Others have negated a role for Tregs in obesity (110, 111). The natural killer T (NKT) cells have also been implicated, with more recent reports suggesting that HFD increases invariant NKT-1 cells (iNKT1) along with proinflammatory ATMs and decreases invariant NKT 10 cells (iNKT10) along with anti-inflammatory ATMs, indicating a role for iNKT cells in the amount of M1 ATMs that invade WAT depots (112, 113). The reader should be aware also of the fact that T cell infiltration in WAT fluctuates with weight cycling between obese states and weight loss states (114). Interestingly, weight loss induces WAT inflammation, which in turn impairs insulin sensitivity by decreasing PI 3-kinase, phosphorylated protein kinase B (PKB), and glucose transporter (GLUT)4 expression (115). In contrast, a long-term ketogenic diet containing extremely high fat causes the depletion of γ/δ T cells, and mice lacking γ/δ T cells show impaired glucose metabolism (116). These results suggested that γ/δ T cells had protective effects on adipose tissue inflammation.

Obesity has been also associated with upregulation of B cells (117), which, although based on limited studies, induces insulin resistance via the activation of T cells and AMTs in obese conditions (117, 118). B cell cytokine production appears to be regulated by activation of the Toll-like receptors (TLRs) in type 2 diabetes inflammation (119).

Mast cells and dendritic cells also accumulate in obese AT (120, 121). Dendric cells promote macrophage infiltration into WAT and liver (122), or maintain T cell homeostasis in WAT (123).

Last, the role of eosinophils has been also explored by Wu et al. (124), who who reported that eosinophils play unique roles in metabolic homeostasis by regulating alternatively activated macrophages.

THE CONTRIBUTION OF ADIPOKINES

Expanding adipocytes themselves produce various mediators in an autocrine fashion, that are able to promote or attenuate WAT inflammation (Fig. 2). These mediators have both immunomodulating and metabolic functions and are collectively termed adipokines. Although the functions of adipokines and their cross-talk are not fully understood, adipokines have emerged as key regulators of WAT metabolic homeostasis and perhaps systemic inflammation. Adipokines secreted from adipocytes not only contribute to localized WAT inflammation (125), but can also induce systemic elevation of cytokines responsible for peripheral insulin resistance (126, 127). Adipokines also facilitate infiltration of macrophage in WAT via IRF7-mediated upregulation of MCP-1 (128) and its receptor C-C chemokine receptor 2 (CCR2) (129–133). An important example of this paradigm is offered by chemerin, an adipokine associated with obesity, inflammation, angiogenesis, and metabolic syndrome. Chemerin regulates leukocyte recruitment into inflamed organs (134–136); adipose depots expressing low level of chemerin maintain normal insulin signaling and a noninflammatory phenotype (137, 138). Calprotectin has been elucidated as another adipokine upregulated in obesity, implicated in WAT inflammation via enhancing adhesion of circulating monocytes or recruitment of macrophages (139–141).

Along with proinflammatory cytokines, adipocytes also secrete anti-inflammatory cytokines, whose abundance and secretion appear to be reduced with weight gain since obesity definitively skews the balance in favor of inflammatory adipokines. Adiponectin is considered the most important of these mediators, with several seminal findings by our laboratory and others’ elucidating its anti-inflammatory actions (142). Silencing of adiponectin exacerbates insulin sensitivity (143, 144). Of note, the expression of adiponectin in WAT is suppressed in acute inflammatory conditions as well as in chronic obesity (145, 146), whereas its overexpression reduces WAT inflammation and improves insulin sensitivity (31). Several molecules have been involved in the regulation of adiponectin expression, which may represent targets of anti-inflammatory drug therapy. Thus, treatment of adipocytes with glucagon-like peptide-1 (GLP-1) analogs increases adiponectin expression (147). Estrogen also increases adiponectin expression and attenuates WAT inflammation (148, 149), which protects from insulin resistance. Notably, a recent study reported that women with metabolic syndrome showed lower adiponectin and IL-6 levels compared with men with metabolic syndrome (150), suggesting that there were sex-specific pathways regulating obesity-induced inflammation. Fat-specific deletion of Fsp27, a lipid droplet-associated protein expressed in adipocytes in mice causes attenuation of WAT inflammation via upregulation of adiponectin (151). However, the Fsp27-null mice model remains insulin resistant, possibly because of hepatic steatosis (151), which underscores the complexity of multiorgan integrated functions such as the metaflammation of obesity and the challenge of interpretation of genetic mouse models.

Taken together, these data suggested the existence of a vicious cycle between dysfunctional adipocytes and the immune system that helps perpetuate low-grade chronic inflammation in the obese organism. Time course studies aimed at dissecting the temporal relationship between activation of relevant WAT cell types, including those found in the stromal vascular fraction, during feeding studies may provide important mechanistic information to help resolve the pathophysiology conundrum of WAT inflammation by uncovering potential therapeutic targets and intervention timing.

MOLECULAR MECHANISMS REGULATING ADIPOCYTE INFLAMMATION

Many molecular mechanisms operating within adipocytes have been suggested as possible regulators of WAT inflammation, including ER stress, hypoxia, and cellular senescence (Fig. 2). With longstanding knowledge linking ER stress with the insulin resistance of obesity and diabetes (152, 153), several recent studies have explored this field and uncovered a number of important data. In mice, a HFD induces ER stress and chronic inflammation in WAT via a FFA-mediated increase of reactive oxygen species (ROS) (154). A more direct causal role for ER stress was demonstrated with genetic manipulation of the ER chaperone 78-kDa glucose-regulated protein (GRP78), an essential ER chaperone and a master regulator of ER homeostasis that operates as a repressor of the unfolded protein response (UPR) stress sensors. Macrophage-selective ER chaperone GRP78 knockout mice are protected from HFD-induced AT inflammation and insulin resistance (155). In addition, ER stress preconditioning protects against FFA-induced adipocyte inflammation through inhibition of the IKK/NF-κB pathway via X-box binding protein-1 (XBP1), a modulator of the UPR (156). Interestingly, obese, insulin-resistant humans also experience higher levels of circulating saturated fatty acid (FFAs) compared with lean individuals (157–160). Furthermore, in obese patients, adipocyte markers of ER stress significantly correlate with BMI or percent body fat (161), and a recent clinical study reported that the amount and types of FFAs in cell membrane phospholipids are related to the inflammatory phenotype of ATMs; for instance, the proportion of proinflammatory ATMs increased with the proportions of palmitic or palmitoleic acids (162). Conversely, omega-3 fatty acids have been shown to exert protective effects on WAT inflammation via G protein-coupled receptor 120 (GPR120), an omega-3 fatty acid receptor (163).

Metabolism of FFA is mainly regulated by cell surface-expressed lipases and cytosolic FABP4. In mice, WAT deletion of the hormone-sensitive lipase (HSL) that regulates mobilization of stored fat induces WAT dysfunction and fatty liver (164), which is consistent with clinical observations showing that null mutation in the HSL gene is associated with WAT inflammation or increased risk of T2DM (165). FABP4 is secreted from adipocytes, where it controls lipolysis, and circulating FABP4 is elevated in metabolic disorders. In vitro treatment of adipocytes with FABP4 inhibits adipocyte differentiation and instigates adipocyte inflammation via the p38/NF-κB pathway (166). In a similar line of research, Koliwad et al. (167) reported that overexpression of diacylglycerol O-acyltransferase-1 (DGAT1) in adipocytes, the enzyme that catalyzes the conversion of diacylglycerol and FFAs to triglyceride (TG), attenuates AT inflammation with improved insulin sensitivity and glucose tolerance in HFD-fed mice, despite the exacerbated accumulation of TGs in liver or skeletal muscle. Hematopoietic deletion of the fatty acid translocase and the scavenger receptor CD36 that takes up long-chain fatty acids (FA) also attenuates WAT inflammation (168).

Research has focused on TLR4 in the inflammatory action of FFAs in obesity. Earlier reports suggested that FFAs serve as chemokine inducers to initiate macrophage infiltration via IKKβ and JNK pathways, and this effect of FFAs is partially through TLR4 (169). More recently, it has been reported that FFAs more directly stimulate TLR4 and initiate TLR4-dependent gene expression of lipid metabolic pathways and membrane lipid composition that promote the inflammatory cascade (170, 171). Furthermore, the expression level of tenascin C, an endogenous activator of TLR4, is also increased in obese WAT (172, 173). Accordingly, TLR4 mutant mice exhibit reduced AT inflammation, preserved adiponectin expression, and increased insulin sensitivity despite higher epididymal AT mass and adipocyte hypertrophy (174). Furthermore, global TLR4 silencing protects mice from AT inflammation, probably via reduced ER stress, augmented autophagy activity, attenuated senescence (175), and reduced oxidative stress through metabolic reprogramming of mitochondria (176). On the other hand, TLR4 silencing in myeloid cells alone does not preserve insulin sensitivity, indicating the importance of TLR4 expression in metabolic organs in obesity (177, 178) At the mechanistic level, the high-mobility group box protein-1 (HMGB1) initiates TLR4-dependent inflammation via NF-κB and p38 MAPK signaling with subsequent IL-6 secretion (179). TLR4-mediated macrophage activation (180, 181) has also been associated with alterations of gut microbiota due to changes in gut permeability following intake of fat diets (180–182). Similar to what we have discussed for cytokines, counterregulatory expression of other TLR types appears to take place in obesity. Thus, TLR9 expressed in adipocytes was found to have WAT anti-inflammatory actions, as demonstrated by significantly reduced adiponectin secretion and increased MCP1 secretion in TLR9 knockout mice (183).

ADAPTIVE REACTIONS INDUCED BY AT INFLAMMATION

Chronic low-grade inflammation mediates not only metabolic dysfunctions but also the adaptive reactions necessary to maintain physiological conditions in WAT. Interestingly, many of these adaptive reactions are inflammatory in nature. For instance, cellular senescence is another contributor to WAT inflammation, and adipocyte death is enhanced in obese mice and humans (24, 184). Furthermore, accumulation of senescent T cells and endothelial cells occurs in the obese WAT (185–187), as well as in the insulin-resistant organism (185, 188). Upregulation of senescence markers such as p16, EGFP, and β-galactosidase has been reported in expanding VAT depots (185). Since senescent cells are known to secrete proinflammatory cytokines (189), it is quite likely that cellular senescence instigates WAT inflammation in obesity. In support of this view, mice lacking programmed cell death-4 (PDCD4), a selective protein translation inhibitor, are protected from HFD-induced obesity and WAT inflammation (190).

With expansion, AT also becomes hypoxic due to insufficient vascular supply (reviewed in Ref. 191). The reduced oxygenation of obese fat depots remains even in the face of compensatory angiogenesis (192, 193). The ensuing hypoxia contributes to the increased secretion of inflammatory adipokines, decreased expression of adiponectin, and upregulation of hypoxia-inducible factor 1α (HIF-1α) (194–196). HIF-1α increases macrophage infiltration and WAT inflammation (197). Relevant to what we discussed in the section above, Snodgrass et al. (198) reported that FFAs exacerbated macrophage-mediated inflammation in response to HIF-1α. Indeed, HIF-1α favors the M1 ATM switch by increasing saturated fatty acid stimulation of the adenine nucleotide translocase-2 (ANT2) (199), and activation of the interleukin-1 receptor-associated kinase M (200). In addition to macrophages, infiltration of other myeloid cells is also mediated by HIF-1α (201).

Hypoxia also negatively impacts adipocyte homeostasis. Thus, hypoxia enhances adipocytokine promoter hypomethylation via the HIF-1α and ten-eleven translocation-1 (TET1) pathway in human adipocytes (202). Taken together, hypoxia in expanded adipocytes is one of the major regulatory mechanisms regulating WAT inflammation and insulin resistance (203).

Adaptive changes have also been described in the lymphatic system of WAT. The lymphatic system is involved in adipose metabolism by regulating lipid absorption and lipid transport and the trafficking of immune cells. HFD-fed obese mice have impaired lymphatic vessel function (204, 205). Clinical studies also found evidence of lymphatic vessel dysfunction in humans (206, 207). Obesity-induced lymphatic dysfunction appears to be related to perilymphatic inflammation (208–210) and is causative of WAT and fibrosis (211) via increased lymphatic system permeability (212). Recent publications have linked adipokines such as leptin and adiponectin to lymphatic function. Leptin causes lymphedema in obese patients via morphological change of lymphatic ducts (213). Importantly, adiponectin can improve lymphedema in obese mice by promoting lymphatic formation via AMPK/ALT/eNOS signaling in lymphatic endothelial cells (214).

INFLAMMATION AS A THERAPEUTIC TARGET IN OBESITY

Attenuation of WAT inflammation in obesity has been explored in preclinical and clinical studies as a possible treatment strategy to avert the metabolic and vascular complication of obesity. Unfortunately, the data obtained have yielded controversial results and have also made clear the difficulty of translating results obtained in animal models of obesity to humans.

Clinical reports have investigated anticytokine therapy in the fight against insulin resistance, based on evidence of elevated TNFα in the serum of obese patients. The mechanistic value of this correlative association has been confirmed with studies demonstrating that administration of TNFα to laboratory animals and humans impairs insulin sensitivity (9, 128). In line with these results, some human studies have reported reduced glucose intolerance in overweight individuals and reduced progression to T2D in patients with rheumatoid arthritis following TNFα neutralization (132, 133). To the contrary, others have reported lack of improved insulin sensitivity in people with T2D patients (129, 130) or rheumatoid arthritis (131) following anti-TNFα therapy. Discrepant results have been reported also in healthy volunteers in whom TNFα infusion increased glucose uptake in skeletal muscle (134). Recent evidence in humans have also implicated IL-1β. However, in the recent CANTOS trial, IL-1β inhibition with canakinumab failed to reduce HbA_1c, glucose or insulin levels in diabetic patients (150). This dichotomy between WAT inflammation and metabolism is also observed at the cellular level and in response to diet and lifestyle intervention and even bariatric surgery. Although it is well established that obesity causes infiltration of ATMs in WAT, calorie restriction, or rapid weight loss, which improve insulin sensitivity, paradoxically it increases the number of ATMs or proinflammatory cytokines both in animal models (215–217) and human obesity (75, 218–222). Thus, despite improvement of systemic metabolic functions, WAT inflammation or the infiltration of resident macrophages are not changed after weight loss (217, 223). Similarly, the expression of proinflammatory genes in WAT is unchanged in the early phase of a very-low-calorie diet but increases in the later phase (224). Instead, high-intensity exercise improves macrophage polarization toward the anti-inflammatory phenotypes (225). Calorie restriction accompanied by eccentric exercise is also reported to upregulate anti-inflammatory macrophages and downregulate proinflammatory macrophages (226). To complicate matters, short-term overfeeding to healthy normal or overweight individuals did not cause upregulation of proinflammatory gene expression or macrophage infiltration in SAT despite the increased body weight and reduction in insulin sensitivity (227, 228). What appears to be important in reducing ATM infiltration and WAT tissue inflammation is the duration of weight loss more than its total reduction (218, 229–231). Future studies should focus on understanding how chronic adaptation to weight gain and weight cycling links WAT inflammation to insulin resistance. More importantly, integrative physiological studies should be undertaken to fully evaluate the interplay between diet-induced WAT inflammation and the functions of other metabolic organs such as the cardiovascular system, skeletal muscle, and liver.

Many researchers have also investigated the impact of rapid weight loss with bariatric surgery on WAT inflammation, but the results are not consistent. Some studies reported the reduction of WAT inflammation with improvement of insulin sensitivity after bariatric surgery (230, 232, 233). Conversely, other studies show no or only minor changes in AT inflammation after surgery (219, 220, 234). Schmitz et al. (235) reported that improved inflammation and insulin action in liver could improve systemic glucose homeostasis such as insulin sensitivity after bariatric surgeries, even if the WAT inflammation parameters, including the number of ATMs, were not significantly affected. Others have reported increased neutrophil infiltration in WAT after bariatric surgery (219, 220).

Several pharmacological and genetic modification strategies have also been investigated in the fight against obesity and metaflammation. In this context, melatonin, a potent antioxidant that improves inflammatory responses and energy metabolism, has been shown to attenuate obesity-associated complications via decreasing inflammatory adipokines including IL-6, MCP-1, leptin, and TNFα (236, 237). Inhibition of glycogen synthase kinase-3 (GSK3), a protein-serine kinase that lies downstream of multiple cell-signaling pathways, also reduces WAT inflammation, by suppressing migration of monocytes, favoring apoptosis of macrophages, inactivating STAT3, and reducing the levels of FFAs or chemokines secreted by VAT (238).

Genetic modifications have been performed in several mouse models in which the link between WAT inflammation and insulin resistance has been carefully investigated. For example, mice overexpressing calpastatin, the endogenous inhibitor of the calcium-dependent protease calpain, show reduced adipocyte apoptosis and macrophage infiltration in WAT (239). Deletion of C1q/TNF-related protein-7 (CTRP7), a secretory protein of the C1q family, in mice is also improved WAT inflammation and insulin resistance, probably via decreased oxidative and ER stress (240).

CONCLUSION AND FUTURE DIRECTION

In this this review article, we have attempted to summarize our current understanding of the role that WAT inflammation plays in the metabolic complications of obesity by highlighting the most recent preclinical and clinical studies. The past decade has clearly demonstrated the existence of a strong relationship between WAT inflammation and insulin resistance, glucose intolerance, type 2 diabetes, and obesity. However, numerous knowledge gaps continue to limit our understanding of the complex interactions among adipokines and systemic pathophysiology. Investigations in WAT inflammation have several difficulties to overcome. First, other organs, such as brain, liver, and skeletal muscles, also contribute to metabolic dysfunction, and an improvement of WAT inflammation is usually associated with an improvement of other organ function that also impact on insulin sensitivity. These integrative metabolic responses make difficult, if not almost impossible, to isolate the mechanistic contribution of WAT inflammation to obesity complications in intact physiological systems. Second, a current limitation of the research in the field is that the fragility of adipocytes makes it difficult to preserve the integrity of the tissue under study. Thus, our current models of the basic mechanisms governing WAT remodeling in obesity are mostly based on extrapolations from in vitro or ex vivo biochemical approaches or conventional histological analysis. Although these approaches remain useful, they yield limited translational value, as they are inadequate to study integrated physiological responses. Obviously, our understanding of these phenomena would greatly benefit from the availability of tools enabling in vivo analysis. Third, there are substantial differences in characteristics and cell surface markers of immune cells between rodents and humans, and therefore careful interpretation of data obtained in animal models of obesity are warranted. More importantly, genetically modified mice should be validated with the use of wild-type mice in which changes in the target gene byproducts should be observed following induction of obesity. Finally, clinical studies have a limited view of WAT tissue inflammation due to the fact that they are largely based on omental fat pad because of accessibility. Since fat depots in different regions of the body assume diverse inflammatory phenotypes, inclusion of additional adipose depots in future studies may provide a more comprehensive view in the field. We hope the current review will stimulate new interest in this field and will promote obesity research.

GRANTS

This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grant DK096521 to R. Scalia and by National Heart, Lung, and Blood Institute Grants HL141108 and HL117724 to M. V. Autieri.

DISCLOSURES

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

AUTHOR CONTRIBUTIONS

T.K. prepared figures; T.K. and R.S. drafted manuscript; M.V.A. and R.S. edited and revised manuscript; M.V.A. and R.S. approved final version of manuscript.