A. W. Ferrante Jr, Naomi Berrie Diabetes Center, Columbia University, 1150 St. Nicholas Ave, New York, NY 10032, USA. (fax: +1 212 851 5331; e-mail: firstname.lastname@example.org).
Obesity induces an inflammation state that is implicated in many clinically important complications, including insulin resistance, diabetes, atherosclerosis and non-alcoholic fatty liver disease. Although the cause and the molecular participants in this process remain incompletely defined, adipose tissue has a central role. Obesity-induced production of pro-inflammatory molecules, typified by TNF-α was recognized more than a dozen years ago, and since then more than two dozen other pro-inflammatory molecules induced by obesity have been identified. More recently a critical role for immune cells, specifically mononuclear phagocytes, in generating the obesity-induced inflammation has been identified. Defining the molecular and cellular components of obesity-induced inflammation offers the potential of identifying therapeutic targets that can ameliorate the complications associated with obesity.
Obesity adversely affects the functioning of many tissues of the body, including the pancreas, liver, skeletal muscle, heart, joints and central nervous system. Clinically the accumulation of adipose tissue contributes to the development of type 2 diabetes mellitus (T2DM), hypertension, hypercholesterolemia, atherosclerosis, nonalcoholic fatty liver disease, gall bladder disease, risk for some cancers, arthritis and Alzheimer’s disease. Whilst it is likely that there are multiple molecular mechanisms linking obesity to its complications, inflammation is a common feature that has been implicated in the pathophysiology of many obesity-associated disorders. Over the past decade efforts to delineate the effects of obesity on individual tissues and systemic physiology have focused attention on key inflammatory signalling molecules, both extracellular and intracellular. More recent studies have identified immune cells and in particular monocytes/macrophages as active participants in obesity-induced inflammation and complications. Obesity increases the numbers and activation state of macrophages in adipose tissue, thereby contributing significantly to obesity-induced adipose tissue inflammation. Defining the molecular functions of macrophages in adipose tissue in both lean and obese states, and identifying the factors that regulate their recruitment and function will provide important insights into adipose tissue inflammation and provide tools to understand the systemic consequences of obesity-induced adipose tissue inflammation.
Obesity-induced adipose tissue inflammation
In the 1960s, population studies found that compared to lean individuals obese subjects have elevated circulating concentrations of systemic markers of inflammation including fibrinogen and other acute phase reactants. Since that time, the circulating concentration of more than a dozen pro-inflammatory, acute phase and pro-coagulant proteins have been identified as being increased by obesity. However, the role of adipose tissue in this inflammatory response was largely unrecognized, until a series of studies published by Hotamisligil, Speigelman and their colleagues. They demonstrated that in rodents and humans obesity increases the adipose tissue expression and secretion of TNFa, a prototypical inflammatory cytokine. Subsequently, the adipose tissue expression of more than a dozen genes that encode inflammatory proteins, including Il-6, MCP-1, iNOS, MMPs and lipocalin, were found to be correlate with adiposity.
In mice, genetic or pharmacological inhibition of TNFa, iNOS, MCP-1 or PAI-1 function in obese mice attenuate obesity-induced insulin resistance. Whilst targeting single genes does not completely protect mice from obesity-induced insulin resistance and whilst efforts to inhibit TNFa in humans have not yielded consistent results, these data implicate inflammation in the development of a clinically important complication of obesity.
In addition to inflammatory cytokines and intercellular signalling molecules, recent data have implicated intracellular pathways that regulate inflammation in the development of obesity-induced insulin resistance. NF-kB is a multi-protein transcription factor whose regulatory targets include secreted inflammatory proteins like TNFa and MCP-1. Many inflammatory stimuli – including activation of Toll-like receptors, reactive oxygen species, ultraviolet radiation and proinflammatory cytokines – lead to the phosphorylation and subsequent degradation of the inhibitory component of NF-kB. Once relieved of its inhibitory component (IKB1 or IKB2) the NF-kB complex is active and translocates to the nucleus where it activates NF-kB-dependent transcription in a cell-type specific manner. leading to its translocation into the nucleus [1–4].
Obesity increases active, nuclear-localized NF-kB in liver and skeletal muscle and transcription of NF-kB target genes [5, 6]. Targeted deletion of I-kappa-B kinase, the kinase primarily responsible for NF-kB activation, in hepatocytes, reduces the obesity-induced hepatic inflammation (e.g. Il-1b, Il-6, SOCS-1, SOCS-3) and the circulating concentration of inflammatory cytokines Il-1b and Il-6 [6, 7]. The reduction in hepatic and systemic inflammation in obese mice with target mutations in Ikbkb also reduces insulin resistance [6, 7]. Pharmacologic inhibition of NF-kB through use of high doses salicylates also improves insulin sensitivity in obese rodent models of insulin resistance and in humans [5, 8].
In a manner similar to its effects on NF-kB, obesity increases the activity of the JNK family of kinases in liver, muscle and adipose tissue . The JNK kinases consists of three structurally related serine/threonine kinases (JNK1, JNK2, JNK3) that are also key regulators of inflammation. JNKs were identified in studies of ultra-violet radiation-induced transcription, but are now known to be activated in response to many cellular stresses and in response to inflammatory cytokines, including TNFα (Rosette, 1996 no. 455; Urano, 2000 no. 457; Liu, 1996 no. 456). Genetic studies identified JNK1 as an important contributor to obesity-related phenotypes. Genetic deletion of JNK1 in mice attenuates the development of obesity in dietary and genetic models of obesity. Deficiency of JNK1 also ameliorates insulin resistance in these rodent obesity models . Impairment of hepatic JNK activity was sufficient to improve insulin sensitivity in obese mice in the absence of effects on weight . Systemic treatment of obese, insulin resistant mice with a peptide that inhibits JNK kinase activity, also reduces insulin resistance in murine models of obesity .
Adipose tissue macrophages
Early studies by Hotamisiligil and subsequent studies identified adipose tissue as an important source of obesity-induced inflammation as measured by the expression of pro-inflammatory molecules . A model emerged in which adipocytes were the primary source and the local target of these inflammatory factors. In this model metabolic stressors, perhaps related to adipocyte hypertrophy, induced an inflammatory response in adipocytes which acted in an autocrine fashion to impair adipocyte function, including insulin signalling . Isolation of stromal vascular cells, i.e. nonadipocytes, from adipose tissue of mice demonstrated that nonadipocytes express significant amounts of the pro-inflammatory molecules produced by adipose tissue . That adipose tissue is infiltrated by immune cells with the onset of obesity had not been suggested until two paper found that obesity increases the macrophage content of adipose tissue in obese states [15, 16]. In mice mesenteric, perigonadal, perirenal and subcutaneous adipose tissue depots the percentage of macrophages correlates positively with measures of adiposity, including body mass index, per cent body fat and adipocyte size. A similar relationship is found between macrophage content and adiposity measures in human subcutaneous adipose tissue. Initial studies demonstrated that the cells are bone marrow derived and dependent on the macrophage-specific growth factor, CSF-1 (also known as M-CSF) . When fractionated using fluorescent activated cell sorting, the macrophages express the majority of several of the key pro-inflammatory molecules, including TNFa, Il-1b, iNOS when compared with adipocytes or nonmacrophage stromal vascular cells . These initial observations suggested that the inflammatory response of adipose tissue to obesity was more complex than anticipated and included the participation of mononuclear phagocytes in a dialogue with adipocytes.
The metabolic, endocrine and inflammatory profile of adipose tissue is depot dependent. On a per gram basis, subcutaneous depots express significantly more leptin and less TNFa than visceral depots . Consistent with macrophage contributing substantially to the inflammatory profile of adipose tissue, the macrophage content of subcutaneous adipose tissue depots is reduced compared with visceral depots in both mice  and humans .
Adipose tissue inflammatory gene expression is dynamic, modulated by weight gain and loss, and by insulin sensitizing therapies. Similarly the macrophage content of adipose tissue is dynamic. In morbidly obese humans subject weight loss following bariatric surgery reduces the adipose tissue macrophage content of subcutaneous adipose tissue in a manner that parallels weight loss . Reduction in inflammatory gene expression following weight loss was measurable for several inflammatory genes, eg Il-6, MCP-1, whose expression was derived primarily from macrophages . Thiazolidinediones are powerful insulin sensitizing agents that also possess potent anti-inflammatory properties toward macrophages . Treatment of mice and humans with thiazolidinediones reduces ATM content whilst concomitantly reducing inflammatory gene expression and improving insulin sensitivity [16, 21]. In contrast, metformin, another clinically important insulin sensitizing agent acts, at least in part, through a hepatic AMP kinase-dependent pathway, and does not lower ATM content of treated humans subjects . Hence, ATM content is dynamically determined, affected both by changes in weight and insulin sensitivity.
Elicited macrophages are bone marrow derived and arise from circulating monocyte progenitors . Monocyte chemoattractant proteins (MCPs) are structurally related member of the chemokine family of proteins and are critical factors in the recruitment of monocytes and elicited macrophages to sites of injury and inflammation . In adipose tissue, the expression of MCP-1 (also known as CCL2) correlates positivitely with adiposity [24–26]. Its expression is higher in visceral compared to subcutaneous depots  and is reduced by treatment with thiazolidinediones [26, 28]. The primary receptor for MCP-1, CCR2, is expressed on circulating monocytes and adipose tissue macrophages. In addition, the expression of two other chemokines that bind CCR2, MCP-2 (CCL8) and MCP-3 (CCL7) are also elevated in adipose tissue of obese mice. Genetic deficiency of CCR2 or MCP-1 reduces macrophage accumulation in adipose tissue of high fat fed obese C57BL/6J mice and partially protects against the development of obesity-induced inflammation, hepatic steatosis and insulin resistance [28–30]. Similarly, pharmacological antagoinism of CCR2 or impairment of MCP-1 binding to CCR2 similarly ameliorates the development of insulin resistance in obese mice [28, 30]. On the other hand, transgenic expression of MCP-1 increases macrophage accumulation of in adipose tissue and impairs insulin’s ability to suppress hepatic gluconeogenesis [28–30].
The phenotypes of CCR2 and MCP-1 deficient mice are not identical, nor does deficiency of either completely reverse the metabolic consequences of obesity [28–30]. Receptor deficiency in addition to ameliorating the metabolic and inflammatory consequences of obesity, also partially protects C57BL/6J mice from the development of obesity when fed a high-fat diet. MCP-1 deficiency does not. The reduction in macrophage content in adipose tissue is ∼30%. This difference suggests that although MCP-1 may a dominant factor in recruitment of monocytes to adipose tissue other chemokines that bind CCR2 also contribute and participate in regulation of feeding behaviour. The effects of CCR2 are also strain dependent. When fed a high fat diet DBA1/J mice deficient in CCR2 were equally obese and insulin resistant as their wild-type littermates. The response of mice to inflammatory signals and metabolic stressors, e.g. high-fat diet, is strain-dependent. Together these data suggest that MCP-1/CCR2 signalling place a role in the recruitment of macrophages to adipose tissue and the associated inflammatory and metabolic consequences, but that other factors clear participate and the relative contribution of these factors is dependent on genetic background.
The adhesion of monocytes to activated endothelium is the initial step in the egress of monocytes from the circulation and the accumulation of elicited macrophages at sites of inflammation. The adipose tissue expression of adhesion molecules, include integrins ICAM-1 and VCAM-1, is increased in obesity [15, 31]. A study of mice fed a Western style diet found no effect of Icam1 genotype on macrophage content of adipose tissue, although it is not clear to what degree the mice developed obesity. There are also conflicting data on the metabolic effects of deficiency of ICAM-1 and MAC-1; some studies have suggested that target deletion of these genes in mice leads to obesity [32, 33] whilst other have found little effect . Identifying the key, molecules required for monocyte adhesion and subsequent macrophage accumulation will require further genetic and pharmacological studies.
Macrophages are remarkable versatile cells that play key roles in maintaining the proper functioning of tissue through the body. They serve general roles of clearing cellular debris, recognizing and initiating an immune response to foreign pathogens, repairing tissues in response to injury and playing a pivotal role in the resolution of inflammatory responses. Macrophages also serve tissue-specific functions. For example, osteoclasts reside within bone and are responsible for bone resorbtion that is necessary to maintain healthy bone. Whilst adipose tissue macrophages undoubtedly play the function of adipose tissue macrophages in obese states is not.
The differentiation of monocytes into mature macrophages is regulated by colony stimulating factors (CSFs). CSF-1 (also known as M-CSF) is the primary regulator of macrophage and dendritic cell differentiation . Adipocytes produce CSF-1 and The spontaneous mutation osteopetrosis (Csf1op/op) is a non-sense allele of the Csf1 gene, and mice that homozygous for this mutation have severe osteopetrosis, due to an inhibition of osteoclast development. Mice lacking CSF-1 are also infertile, have impaired hypothalamic-gonadal axis development, lack teeth, are smaller in size than wild-type mice, have reduced adiposity, are blind and are protected from atherosclerosis. A detailed study of their metabolic function has not been undertaken but interpretation any results would be difficult given the complicated developmental abnormalities from which CSF-1 deficient mice suffer. Mice which lack the receptor for CSF-1, Csf1r−/−, have an nearly identical phenotype. Thus a better understanding of the role of macrophage maturation in adipose tissue awaits development of a targeted deletion approach, i.e. deletion of Csf1 in adipocytes or the receptor in macrophage of adipose tissue.
The production of immune modulatory molecules, including pro-inflamamtory TNFa, Il-6 and Il-1b, by macrophages is a characteristic of classical activation . Resident macrophages in their role as sentinels exist in a quiescent state, producing low amounts of immune modulator molecules. However, when a macrophage encounters and recognizes a foreign pathogen and is stimulated by a T helper cell type 1 (TH1) lymphocyte through action of TH1 cytokines, e.g. interferon-gamma, macrophages are activated and in an antigen-independent fashion gain greater anti-microbial activity. In addition, they produce proinflammatory molecules that amplify and help coordinate the antimicrobial response . The production of many proinflammatory cytokines by adipose tissue macrophages suggests that they are indeed activated, but there is scant evidence for microbial-dependent process . Furthermore, the levels of expression of inflammatory cytokines on a per cell basis is likely at least an order of magnitude lower than what is induced by a classical microbial stimulator such as lipopolysaccharide. Defining the factors that activate adipose tissue macrophages will likely identify fundamental aspects of adipose tissue that are altered by obesity.
A primary function of macrophages is the clearance of dead cells, both necrotic and apoptotic. When cells undergo apoptosis alterations in cellular membranes are recognized by receptors on macrophage. Clearance of apoptotic cells by macrophages does not induce classical activation. Cinti and colleagues have suggested that obesity induces adipocyte necrosis, perhaps secondary to hypertrophy, and that necrosis, unlike apopotosis, would explain the inflammatory response of adipose tissue macrophages to obesity . However, the induction of adipocyte apoptosis in a transgenic model of inducible lipoatrophy also leads to the accumulation of macrophages in adipose tissue and modest inflammatory response . Hence, the relative contribution of adipocyte necrosis and apopotosis to macrophage accumulation in adipose tissue await further elucidation. It is likely that in addition to cell death, other factors e.g. elevated local concentrations of fatty acids, tissue hypoxia and hypertrophy induced production of chemoattractants play a role in macrophage recruitment, maturation and turnover. Identifying the molecular signals that regulate these processes will provide insights into the pathophysiology of obesity.
The recently discovered dynamic nature of adipose tissue macrophage has offered exciting new insights into the physiology of adipose tissue. An chemalin overview of the putative role of macrophages in adipose is depicted in Fig. 1. Accumulating evidence has documented that adipose tissue macrophage content and activation are strongly associated with obesity-induced complications, both in rodents and humans. The association is provocative and some data in mice implicate these immune cells causatively in complications such as insulin resistance, hepatic steatosis and dyslipidemia. The study of these cells is still early and most questions about their function remain unanswered. Over the next decade, investigators will likely better define key aspects of immune cell function in adipose tissue in particular: (i) identifying the mechanisms that recruit, retain and activate mononuclear phagocytes populations; (ii) defining the local functions of adipose tissue macrophages; and (iii) Determining the systemic consequences of adipose tissue macrophage function and dysfunction. Knowledge of these basic processes will greatly inform our understanding of adipose tissue biology and identify potential targets for therapeutic intervention.
Conflict of interest statement
No conflict of interest was declared.
This work described herein was supported by grants from the NIH (DK066525, DK59960, the Columbia Diabetes & Endocrinology Research Center – DK063608, and the Berrie Foundation).