Metabolic Syndrome, Insulin Resistance and Roles of Inflammation-Mechanisms and Therapeutic Targets

Abstract

Obesity and its co-morbidities, including type 2 diabetes (T2D) and cardiovascular disease (CVD), are associated with a state of chronic low-grade inflammation that can be detected both systemically and within specific tissues. Areas of active investigation focus on the molecular bases of ‘metabolic’ inflammation and potential pathogenic roles in insulin resistance, T2D and CVD. An increased accumulation of macrophages occurring in obese adipose tissue has emerged as a key process in ‘metabolic’ inflammation. Recent studies have also begun to unravel the heterogeneity of adipose tissue macrophages, and their physical and functional interactions with adipocytes, endothelial cells and other immune cells within the adipose tissue microenvironment. Translating the information gathered in experimental models of insulin resistance and T2D into meaningful therapeutic interventions is a tantalizing goal with long-term global health implications. In this context, ongoing clinical studies are testing the effects of targeting inflammation systemically on metabolic and cardiovascular outcomes.

Excess adiposity increases risk for developing a variety of pathological conditions, including T2D^1,2, CVD³, steatohepatitis, asthma⁴, and several types of cancer⁵. Mechanistic studies suggest that a state of chronic subacute inflammation may promote the onset and modify the severity of each of these diseases, thus representing a potentially unifying pathogenic link.

Obesity induces a low-grade inflammatory response

The chronic, subacute inflammatory state that accompanies obesity is evident both systemically and more focally in affected tissues, including adipose tissue, liver and the vasculature. Moreover, the inflammatory changes associated with obesity can be found in both immune cells and non-immune, parenchymal cells within these tissues, and, in common with classical forms of acute inflammation, include the abnormal production of cytokines and chemokines that may further attract and activate immune cells⁶. In keeping with its more indolent and chronic nature, obesity-associated inflammation elicits changes of much smaller magnitude and is not accompanied by the cardinal signs of acute inflammation, those being rubor et tumor cum calore et dolore (redness and swelling with heat and pain).

Consistent with a potential role for inflammation, in obese subjects there are increases in both numbers⁷ and activation states⁸ of peripheral blood mononuclear cells, as well as elevated serum levels of pro-inflammatory cytokines (reviewed in 9)⁹. Moreover, large-scale prospective studies have demonstrated that markers of inflammation in aggregate predict incident T2D in Caucasian populations^{10, 11}. Although epidemiological studies are inherently correlative, these and related studies provide conceptual frameworks for addressing such questions as: 1) Does obesity-induced ‘metabolic’ inflammation promote or enhance insulin resistance, and 2) What organs, tissues and cell types are primarily involved?

White adipose tissue (WAT), particularly the visceral form, has been implicated. Early studies by Hotamisligil and Spiegelman postulated that the enhanced production of TNF-α by adipocytes in obese rodents would induce systemic insulin resistance¹² in a cell-autonomous fashion. Additional cytokines, which collectively have been referred to as ‘adipokines,’ were also related to glucose homeostasis and inflammation (reviewed in 13)¹³. Of note, the decrease of anti-inflammatory adipokines such as adiponectin and adipsin may also contribute to adipose tissue inflammation.

However, these hypotheses predated the discovery of macrophages and other leukocytes in the WAT of obese animals and subjects, which are major sources of pro-inflammatory cytokines produced by WAT. Specifically, the abundance of macrophages in the stromal vascular fraction (SVF) of WAT increases as obesity progresses in both humans¹⁴ and rodents^{14, 15}. Correlations between adipose tissue macrophage (ATM) number and body mass¹⁶ suggest that macrophages and associated inflammation might play pathogenic roles in obesity-induced insulin resistance, an issue that is still being debated.

In addition to identifying a new adipose tissue resident cell type, these studies prompted exciting avenues of investigation. This review will focus on i) the diversity of ATM subsets, ii) ATM communication with other adipose tissue cells, and iii) the potential roles of ATMs and inflammation in insulin resistance and T2D.

ATM heterogeneity and plasticity

Macrophages are highly plastic and influenced by local microenvironment. Therefore, before considering ATM macrophage heterogeneity per se, it is valuable to highlight that macrophage heterogeneity is also a result of their environment. Bone marrow-derived, circulating monocytes give rise to tissue-resident macrophages as well as such specialized cells as bone-forming osteoclasts and antigen-presenting dentritic cells (DCs). Thus, microglia in the central nervous system, Langerhans cells in the epidermis, Kupffer cells in the liver, serosal macrophages in the peritoneal space, alveolar macrophages in the lungs, and white and red pulp macrophages in the spleen are all quite distinct in terms of appearance and function, including many expressed genes and proteins^{17, 18}.

Following the T helper cell Th1/Th2 functional categorization, macrophages can be grouped into polarization extremes according to activation state (reviewed in^{17, 18}). The M1 or ‘classically’ activated macrophages are produced in response to bacterial lipopolysaccharide (LPS) or interferon-γ (IFN-γ), and in turn produce pro-inflammatory cytokines (e.g. IL-1β), exhibit anti-bacterial host defense capabilities¹⁹, and promote a Th1 lymphocyte response. In contrast, M2 or ‘alternatively’ activated macrophages are formed in response to IL-4 or IL-13 treatment. These have been referred to as M2a macrophages, which help clear parasites²⁰ and express a high ratio of arginase to inducible nitric oxide synthase (ARG/iNOS), compared to M1 macrophages²¹. M2a macrophages support Th2 responses via IL-10 production. A second subset of M2 macrophages produced in response to IL-10 are referred to as M2c or ‘deactivated’ macrophages²². Thus the M2 category includes a heterogeneous array of non-M1 macrophages with properties ranging from tissue repair to anti-inflammation.

Although the M1/M2 classification system is oversimplified, it provides a useful initial framework for distinguishing macrophage functions. Depletion and reconstitution approaches have also helped to determine that distinct monocyte precursors in the circulation give rise to M1 or M2 macrophages. Littman and colleagues used these procedures to subdivide circulating monocytes into ‘inflammatory’ and ‘resident’ subtypes²³. ‘Inflammatory’ monocytes in mice are short-lived, preferentially recruited to sites of inflammation, and give rise to M1 macrophages. In mice ‘inflammatory, pre-M1 type monocytes’ are distinguished using flow cytometry to identify the cell surface expression of a select set of proteins: Ly6C^hi CCR2⁺ CD62L⁺ CX₃CR1^low. By contrast, the Ly6C^low CCR2⁻ CD62L⁻ CX₃CR1^high monocytes in mice survey tissues for responses to injury or damage²⁴, and migrate to both inflamed and non-inflamed tissues where they differentiate into M2-like macrophages with remodeling and anti-inflammatory functions.

In lean mice, resident ATMs express prototypical M2 markers, including IL-10, Ym1 (chitinase 3), ARG, and the lectin MGL1^{25, 26}. ATMs in lean mice are also diffusely located between adipocytes throughout the fat pads. Lumeng et al., conducted an interesting experiment that aimed to compare newly recruited ATMs in the fat pads of obese mice to tissue resident ATMs in the fat pads of lean mice. To this end, they combined pulse-chase labeling of ATMs with flow-cytometry. Newly recruited macrophages in fat pads of obese mice were MGL1⁻ CCR2⁺, expressed high levels of iNOS and IL-1β, and localized to crown-like structures (CLS). By contrast, MGL1⁺ ATMs in obese mice were more evenly distributed between adipocytes, as also seen in the fat pads of lean mice. Of note, the surface marker CD11c primarily co-localized with newly recruited, MGL1⁻ ATMs in CLS²⁶.

Models suggesting that a phenotypic switch in ATMs accompanies weight gain and causes inflammation-induced insulin resistance are appealing because they are simple, but underestimate ATM heterogeneity in obese WAT. ATMs do not lie at the M1/M2 extremes of macrophage activation, but are more in the middle of the classical activation spectrum. When analyzed for MGL1, SVCs isolated from mice fed high-fat diet (HFD) for eight weeks harbored a subset of newly recruited ATM with intermediate MGL1 expression (MGL1^med)²⁷. In apparent contrast with previous reports, MGL1^med ATMs were also positive for CD11c and accounted for the majority of CD11c⁺ ATM in CLS. Gene expression profiling suggested that MGL1^med CD11c⁺ as well as MGL1⁻ CD11c⁺ have a ‘mixed’ M1/M2 phenotype. After 12 weeks of HFD, at a time when WAT tissue remodeling becomes prominent²⁸, both MGL1⁻ CD11c⁺ and MGL1^med CD11c⁺ subsets expressed high levels of proteins promoting tissue repair, including MMP-12, and M2 signature proteins such as ARG and Ym-1. It has also been shown that MGL1, which identifies ATMs in lean mice, is required for accumulation of CD11c⁺ ATMs in obesity²⁹. These studies illustrate the heterogeneity, plasticity and partial overlaps between ATM phenotypes, and suggest that CD11c⁺ ATMs can also participate in tissue repair as opposed to exclusively promoting inflammation and insulin resistance³⁰.

Consistent with the picture emerging in mouse models, ATMs in human obesity also appear to have an M2 bias. Expression of MMPs, CD209 (DC-sign), and other M2 markers are up-regulated in CD14⁺ CD16⁻ ATMs, especially those that are CD14⁺ CD206^{hi 31, 32}. However, the human samples for such studies are often obtained from morbidly obese subjects, presumably at late stages in their ’natural history’ that might resemble the remodeling changes observed in mice after longer times on HFD²⁸.

WAT leukocytes: the art of networking in a busy neighborhood

In addition to ATMs, other immune cells found in WAT include CD4⁺ and CD8⁺ T cells, NKT cells³³, B cells³⁴, eosinophils³⁵, neutrophils³⁶, and mast cells³⁷. Flow-cytometry and gene expression analyses and imaging methods have been used to show that T cells are present in WAT and increased in rodent and human obesity. Compelling evidence has defined discrete and partially opposing regulatory functions for CD4⁺ and CD8⁺ cells in obesity-associated WAT inflammation and IR^38–40. Nishimura et al. showed that obesity increases CD8⁺ number (even when normalized for WAT weight), and that CD8⁺ infiltration precedes and promotes HFD-induced ATM accumulation. In addition, CD8-deficient mice exhibited improved insulin sensitivity.

Two additional reports investigated the role of CD4⁺ regulatory T cells (T_reg), which can suppress immune responses by, among other mechanisms, coaxing macrophage differentiation toward an anti-inflammatory M2 phenotype. Having noted reduced numbers of WAT T_reg in both genetic (ob/ob) and induced models of obesity (HFD feeding), Feuerer et al., tested the consequences of both T_reg ablation and gain-of function on insulin sensitivity⁴⁰. Deletion of Foxp3⁺ T_reg resulted in impaired insulin signaling in liver and WAT and marked up-regulation of WAT inflammatory cytokines, whereas boosting T_reg function using IL2-based complexes partially protected against the development of HFD-induced insulin resistance. In human adipose tissue samples corresponding inverse correlations between BMI and the T_reg marker Foxp3 suggest that T_reg may play roles in human obesity as well. Winer et al. corroborated these findings by showing that reconstitution with CD4⁺ cells but not CD8⁺ improved the metabolic phenotype of Rag1-null mice, which are deficient in B and T lymphocytes and exhibit accelerated IR³⁹.

Finally, WAT T_reg seem to express a discrete T cell receptor repertoire⁴⁰ suggesting that there might be a discrete antigen or set of antigens recognized by T_reg and possibly other T cell subtypes (e.g. a modified lipid?).

In summary, a major goal in the field will be to define the hierarchy among immune cell types migrating to WAT in response to obesity. ATMs account for the majority of effector cells in the obese SVF, yet, obesity-associated WAT inflammation could be the consequence of alterations in numbers or activation states of other regulatory cells, including T_reg, that maintain WAT immune homeostasis in the lean state.

Relevance of ATMs in insulin resistance

Although the degrees of adipose tissue inflammation in obesity correlate with the severity of insulin resistance and T2D, this does not prove that ATMs act as pathogenic mediators of these conditions in humans. To address this question in rodents, inflammatory pathways including those controlled by JNK⁴¹ or NF-κB⁴² were manipulated in myeloid cells and the affects on insulin resistance and T2D were assessed. The myeloid activity of the insulin-sensitizing and anti-inflammatory nuclear receptor, PPAR-γ was similarly studied ⁴³.

Inhibition of NF-κB in monocytes/macrophages (as well as DC and neutrophils) was achieved using LysM^Cre-mediated excision of the IκB kinase, IKKβ⁴⁴. This partially protected the mice from developing HFD-induced systemic insulin resistance without significant changes in body weight. With regard to JNK, a strategy of reciprocal bone marrow transplantation (BMT) in Jnk-null and wild type mice was used to distinguish between effects in hematopoietic and non-hematopietic compartments on HFD-induced insulin resistance, inflammation, and adiposity⁴⁵. While chimeric mice lacking Jnk in non-hematopoietic cells were protected from diet-induced obesity (i.e. gained less weight), deletion of Jnk from the myeloid compartment, which includes macrophages, resulted in improved insulin sensitivity and reduced inflammation in both liver and WAT.. Of note, Vallerie et al. showed that the beneficial effects of hematopoietic JNK deletion on insulin sensitivity requires the concomitant lack of JNK in parenchymal cells⁴⁶.

Finally, LysM^Cre-mediated deletion of PPAR-γ led to HFD-induced increase in WAT pro-inflammatory cytokine expression and body weight, and to impaired β-oxidation and insulin sensitivity⁴⁷. These experiments, which were conducted in the Th2-permissive Balb/c background that is biased toward M2 macrophages polarization, showed that PPARγ primes myeloid cells toward an anti-inflammatory phenotype. It should be acknowledged that lack of PPAR-γ in myeloid cells did not result in HFD-induced IR when BMT experiments were performed in the Th1-biased C57BL/6 background⁴⁸.

Neither LysM^Cre-mediated gene deletion nor the bone marrow transplant model selectively target macrophages, and are certainly not selective towards ATMs. Indeed, the characterization of approaches to specifically address the impact of ATMs remains a major need in the field.

This caveat notwithstanding, blockade of prototypical pathways in myeloid cells has been used to support potential roles for the monocyte-macrophage system, and ATMs specifically, in the regulation of insulin sensitivity.

In humans, weight loss in morbidly obese patients who underwent Roux-en-Y bypass procedures resulted in marked reductions of ATMs and CLS in subcutatneous WAT, as well as blunted expression of CCL-2. However, the degree of improvement in insulin sensitivity three months after surgery did not significantly correlate with ATM number, possibly owing to the limited power of the study⁴⁹. Subsequent work from the same group showed that ATM number in omental, but not subcutaneous WAT, correlated with post-surgery insulin sensitivity, but even more significantly with liver inflammation⁵⁰. Interestingly, weight loss in mice is initially associated with the opposite response, as ATMs and other features of inflammation initially increase as opposed to decrease⁵¹. These investigators interpret their findings to suggest that the products of lipolysis associated with weight loss (e.g. non-esterified fatty acids and glycerol), might stimulate the recruitment of monocyte/macrophages into the fat depots.

How should we interpret the differences in the correlation between ATM content and insulin resistance in humans vs. rodents? First, experimental models of obesity are often ‘super-sized’ extremes of the disease that may not capture the whole spectrum of obesity observed in human subjects. Second, human adipose tissue samples are often collected at a later point in the natural history of both obesity and WAT inflammation, when compared to rodent models, at times when active remodeling may already have occurred and ATMs may have become less abundant. This is supported by the significantly lower number of CLS in human obese WAT, when compared to mouse. Finally, glucose homeostasis in humans may be more closely associated with the activation state of ATMs, before and after weight loss, than with the overall ATM number. Recent work has begun addressing the effects of weight loss and improvement in insulin sensitivity on the transcriptome of both adipocytes and ATMs⁵². The comparison of ‘signature’ gene profiles in ATMs from human vs. mouse obese adipose tissue will help further clarify differences and similarities in ATM polarization states occurring in these models.

Targeting inflammation: potential therapeutic approaches in insulin resistance and T2D

Evidence collected in humans and rodents has validated chronic inflammation as a promising target for prevention and therapy of IR, T2D, and cardiovascular disease. Because of its well-established role in inflammation and insulin resistance in animal models, TNFα seemed a rational target for new therapeutic intervention. However, several approaches to antagonize TNFα have had no effect on glucose levels in patients with T2D^{53, 54} and only marginal impact on insulin sensitivity in non-diabetic, insulin-resistant patients^{16, 55}.

In randomized trials with small sample size, the anti-inflammatory drug salsalate was found to curb insulin resistance and inflammatory parameters in obese individuals⁵⁶, and to improve glucose control and triglyceride levels over a 3-month treatment period in patients with T2D¹. A larger, multi-center, double-blind, placebo-controlled NIH-funded clinical trial of 1 year duration was recently completed but the results have not yet been published (TINSAL-T2D; ClinicalTrials.gov registration number: NCT00799643).

The blockade of IL-1R1 by means of a specific binding protein, IL-1RA, improved insulin sensitivity and β-cell secretory profile while reducing markers of systemic inflammation⁵⁷. The beneficial effects on β-cell secretion persisted for >3 years after discontinuation of the IL-1RA. Subsequent clinical trials showed modest glucose-lowering efficacy, which decreased enthusiasm for using IL-1β blockade to treat patients with T2D. However, the results certainly showed that this highly selective immunomodulator lowered blood glucose. Although the magnitude of glucose lowering with IL-1β blockade may be less than one would wish, the fact that both salsalate and IL-1β blockade do indeed lower blood glucose provides strong supporting evidence for roles of inflammation in obesity-induced insulin resistance. Inflammation also plays potentially important roles in the development and progression of atherosclerotic plaque. Since available diabetes drugs have little known impact on atherosclerosis, these new anti-inflammatory approaches may provide a welcome addition to the armamentarium. This will of course need to be tested in an outcomes trial that evaluates all-cause and cardiovascular mortality.

Together, these and other studies suggest that targeting individual prototypical cytokines may not be the most efficacious approach to curb systemic inflammation and to prevent/treat metabolic derangement associated with IR. Also, the results of these trials beg the question whether and to what extent taming inflammation in the adipose tissue accounts for observed systemic anti-inflammatory effects. Since drug provision in human studies is usually systemic, only correlative data can be gathered for now to link effects on adipose tissue inflammation with systemic metabolic improvement.

Concluding remarks

Associations between obesity and several diseases and conditions having inflammatory components have led to hypotheses suggesting that WAT inflammation promotes these co-morbidities. For instance, mediators released by inflamed WAT may exert endocrine effects at the level of the vascular wall and airways, predisposing to atherosclerosis or asthma, respectively. It is still unclear however, which immune cells, primed by trafficking through obese WAT, could elicit effects in other tissues by modifying ‘inflammatory tone.’ This speculative hypothesis implies tissue inter-dependence in the response to anti-inflammatory strategies and the need to assess efficacy at systemic, whole body, levels. Understanding how inflammation arising in one tissue affects the physiology and pathology of other organs remains a tantalizing question with therapeutic implications for chronic conditions including obesity, diabetes, and atherosclerosis.

Abbreviations

CCL-2, also known as MCP-1

C-C motif chemokine ligand-2

CCR2

C-C motif receptor

CXCR1

C-X-C motif chemookine (CXC) receptor-1