Biochemical Journal

Review article

Insulin resistance and atherosclerosis: convergence between metabolic pathways and inflammatory nodes

Robert Stöhr, Massimo Federici


For some time now it has been known that diabetes and atherosclerosis are chronic inflammatory diseases that are closely associated with one another and often develop together. In both there is an increase in tissue-wide inflammation that is exhibited by the infiltration of immune cells into the adipose tissue and the vascular walls respectively. The monocyte/macrophage populations that are recruited in these seemingly different settings also display a high similarity by exhibiting similar phenotypes in both conditions. In the insulin resistant as well as the atherosclerotic setting there is a distinct switch in the macrophage populations present from an anti-inflammatory (M2) population to an inflammatory (M1) population, which releases cytokines and chemotactic factors with the ability to worsen the local environment and thus aggravate the situation by creating a vicious circle. However, although some discoveries suggest that preventing the development of M1 macrophages reduces inflammation and thereby aggravation of these diseases, there are currently no clear-cut opinions on how to achieve a switch from M2 to M1.

  • atherosclerosis
  • cardiovascular disease
  • diabetes
  • inflammation
  • macrophage
  • signal transduction


T2DM (Type 2 diabetes mellitus) is a disease characterized by high blood glucose caused by either tissue insulin resistance or defective insulin secretion and resulting in an impaired biological response at a cellular level [1]. T2DM has become a growing medical concern currently affecting over 285 million people in the world (6.4% of the total adult population) [2]. The most common cause of death in diabetics in Europe is coronary artery disease, with several studies reporting that the risk of macrovascular complications (stroke and myocardial infarction) is 2–4-fold higher in people with diabetes than in non-diabetics [3].

Insulin is able to directly stimulate signal transduction in endothelial cells [4] with both pro- and anti-atherogenic properties having been attributed to it. Pro-atherogenic properties include the activation of endothelin 1 [5] and plasminogen activator inhibitor-1 [6], whereas anti-atherogenic properties are exhibited through decreased endothelial cell apoptosis [7] and increased production of nitric oxide [8]. This system evolved to maintain a physiological ability to vasodilate during physiological hyperinsulinaemia, which is reflected by the paradoxically neutral role of Insr (insulin receptor) deletion specifically in the endothelial cell [9]. However, the metabolic syndrome being composed of other potential pro-atherogenic stimuli, the net effect of defective insulin signalling on the endothelium has only recently been elucidated.

Rask-Madsen et al. [10] showed that conditionally knocking out the Insr gene in vascular endothelial cells of pro-atherogenic ApoE (apolipoprotein E) mice results in a mouse model with the same whole-body insulin sensitivity, glucose tolerance, plasma lipids and blood pressure as the control ApoE. The atherosclerosis-prone animals with defective insulin signalling in endothelial cells, however, developed a larger amount as well as a quicker onset of atherosclerosis. Interestingly, hyperinsulinaemia alone did not increase atherosclerosis in ApoE mice [11].

Although this and many theories have been put forward over the years, the exact molecular mechanism of this obvious link between diabetes and the development of atherosclerosis remain obscure. Two main reasons for the currently still unclear situation appear obvious. First, atherosclerosis is a multifactorial disease involving a wide range of responses of different tissues and cell types to metabolic and environmental stimuli and no single link or mechanism may thus be responsible [12]. Secondly, it appears to be that the cellular response to insulin depends on its binding to a cell surface Insr with an ability to elucidate a different response depending on the cells expressing this receptor. The main action of insulin is to promote glucose uptake mainly in skeletal muscle and adipose tissue [13]. However, it has also been proven that insulin exerts actions completely unrelated to the glucose metabolism, including vessel dilatation and contraction [14] and endothelial dysfunction [15]. The effect of hyperinsulinaemia and insulin resistance may therefore have different effects depending on the tissue affected.

Another part of the human metabolism that is strongly regulated by insulin is lipid metabolism. Diabetics usually exhibit a tendency towards central adiposity with the enlarged visceral adipocytes responding poorly to insulin. This results in inappropriate timing of FFA [non-esterified (‘free’) fatty acid] release causing exposure of non-adipose tissue to excessive fat [16], leading to TAG (triacylglycerol) deposition and thus inflammation in these tissues.

With inflammation being pivotal to the development of atherosclerosis [17] and the monocyte/macrophage axis being involved at every step of the formation of a mature plaque from the fatty streak, it appears that the actions of insulin or its relative absence are both directly and indirectly able to manipulate these events and therefore may explain the increased occurrence of atherosclerosis in diabetics.

Much emphasis in unravelling the mysteries of diabetes and its link to atherosclerosis has been placed on the direct effect of hyperinsulinaemia, hyperglycaemia and insulin resistance on different tissues, including liver, adipose tissue and vascular cells. However, we believe that an emerging, and yet largely not elucidated, field lies in the effect of this diabetic ‘unhappy triad’ to immune modulate the monocyte/macropahge axis to promote inflammation and thus the resulting propensity to develop atherosclerosis.

In the present review we will attempt to summarize recent discoveries regarding the similarities in the changes exhibited at local tissue level involving the monocyte/macrophage axis and the development of atherosclerosis and diabetes.


T2DM and atherosclerosis belong to the group of metabolic diseases driven by chronic low-grade inflammation that not only induces them, but also helps in maintaining and escalating the situation. Indeed the state of insulin resistance, obesity and overt diabetes is characterized by elevated serum levels of inflammation [e.g. IL (interleukin)-6, TNFα (tumour necrosis factor α)], chemokines [MCP-1 (monocyte chemoattractant protein 1), RANTES (regulated upon activation, normal T-cell expressed and secreted)] and acute phase reactants [18,19]. This self-perpetuating state then causes further infiltration of monocytes and macrophages into adipose tissue and the vascular endothelium [20,21] where they promote this cycle (Figure 1). It was first recognized in 2003 by two different groups that the increase in the amount of macrophages present in adipose tissue correlates with weight gain [21,22].

Figure 1 The onset of obesity leads to a generalized state of low-grade inflammation with a general increase in the level of inflammatory markers

These inflammatory markers in turn activate several different pathways in the liver, muscle and adipose tissue and thus result in a vicious circle in which inflammation sustains itself and causes further damage. FA, fatty acid; RANTES, regulated upon activation, normal T-cell expressed and secreted; SFA, saturated fatty acid; TG, TAG; VLDL, very-LDL.

However, the interesting point in this mechanism is that there is not only a difference in the amount, but also the type, of tissue-specific macrophages present between lean and adipose animals [23]. Two major subsets of circulating monocytes have been described in humans and animals on the basis of their expression of superficial cellular markers. In mice, the main differentiator lies in the presence or absence of the surface receptor Ly-6C [24]. The presence of Ly-6C (Ly-6Chi) represents the pro-inflammatory macrophage (M1), which infiltrates early atherosclerotic plaques as well as the adipose tissue of obese animals. In contrast with this, a dominant population of alternatively activated anti-inflammatory macrophages (M2) is found in the adipose tissue of lean animals [23]. Research over the last few years has therefore concentrated on elucidating the differences between these two subpopulations as well as attempting to create a switch from M1 to M2 and thereby potentially breaking the self-perpetuating cycle of local inflammation→macrophage activation→chemokine release→further macrophage activation and increased inflammation (Figure 2).

Figure 2 Effect of obesity on macrophage activation

Adiposity leads to a preferential activation of inflammatory macrophages (M1), which increases tissue-based inflammation through secretion of inflammatory markers such as IFNγ and the release of FFAs. This leads to increased rates of insulin resistance and atherosclerosis development. In lean adipose tissue, alternative activation of macrophages (M2) prevails, thus providing protection from these diseases through the release of anti-inflammatory cytokines such as IL-10.

The M1 macrophage

Hyperglycaemia as well as defective insulin signalling of endothelial cells can promote the adhesion and diapedesis of monocytes into the adipose tissue. These macrophages then become activated by circulating chemokines and begin ingesting retained lipoproteins [25], which causes them to develop lipid droplets within their cytoplasm (known as ‘the foam cell’), which in turn creates an increased inflammatory response through the further release of cytokines [26].

The classically activated M1 macrophage is derived from monocyte activation through IFNγ (interferon γ) being released from CD4+ Th1 cells [27]. It is considered pro-inflammatory, as its activation will result in a population of macrophages with enhanced bactericidal activity as well as increased secretion of pro-inflammatory cytokines such as IL-6 or TNFα.

The M2 macrophage

As already mentioned, a macrophage is not always a macrophage when it comes to the populations present in the adipose tissue, and therefore placing all the blame on it as solely responsible for the state of inflammation is, according to up-to-date discoveries, oversimplified and outdated.

The alternatively activated macrophage, M2, is derived from the same progenitor as the M1 macrophage and there is thus a certain plasticity and ability to shift from one polarization to another in the macrophage. Macrophages polarize towards the M2 state upon stimulation with cytokines IL-4 and IL-13, which are released by Th2 cells. Upon binding of IL-4 or IL-13 to their specific receptors, a cytoplasmic cascade is initiated which culminates in the phosphorylation of STAT (signal transduction and activator of transcription) 6 [27]. Phosphorylated STAT6 is then able to translocate into the nucleus where it induces the expression of its target genes PPAR (peroxisome-proliferator-activated receptor) γ and PPARδ as well as PGC1β (PPARγ co-activator 1β).

Among the genes that are induced in the M2 macrophage is Arg (arginase) 1, which diverts cellular arginine from the production of nitric oxide towards ornithine and polyamine, which are useful in cellular respiration [28].

Alternatively activated macrophages are preferentially dependent upon oxidative metabolism, thanks to their ability to take up and oxidate fatty acids, in response to IL-4 stimulation [29] rather than using glycolysis as an energy source like their M1 counterparts [30]. This STAT6-dependent shift towards oxidative metabolism results in an influx of fatty acids that do not only serve as a source of substrate for β-oxidation, but also activate PPARs resulting in an increase in this process (Figure 3).

Figure 3 The intricate link between T-cell activation and macrophage polarization

Th1 cells are activated through IL-12 release as well as antigen presentation from dendritic cells. The phosphorylation of STAT4 results in increased production of the T-cell transcription factor T-bet1, which in turn increases IFNγ production while inhibiting IL-4 release. IFNγ then, together with other inflammation inducers, such as reactive oxygen species (ROS) and TLR-4, promote the phenotypic activation of inflammatory macrophages (M1) through activation of the JNK and NF-κB pathways. Th2 cells are activated by IL-4, which through STAT6 and the T-cell transcription factor GATA3 results in the increased production of IL-4 and IL-13. These then induce alternative macrophage activation through the induction of PPARγ and PGC1β, resulting in increased β-oxidation of the cell. RAGE, receptor for advanced glycation end-products.


Similar to T2DM, much research has gone into elucidating the role of inflammation in atherosclerosis. The idea that the macrophage and localized inflammation may be involved became apparent when Smith et al. [31] demonstrated that macrophage deficiency severely reduces atherosclerotic burden in the murine ApoE knockout model of hypercholesterolaemia. This idea was furthermore supported by evidence that, in cases of chronic systemic inflammation such as rheumatoid arthritis [32] and systemic lupus erythematosus [33], the incidence of cardiovascular disease rises. As macrophages are able to secrete pro-inflammatory cytokines and have been found to make up to 60% of atheromata mass [34], they are considered critical participants in the atherogenic process.

In both humans and animals the presence of hypercholesterolaemia leads to the accumulation of plasma lipoproteins in the extracellular matrix where they undergo oxidation [35]. The accumulation of these oxidized lipoproteins leads to the formation of early atherosclerotic plaques called fatty streaks. These in turn lead to local endothelial dysfunction resulting in the expression of ICAM-1 (intercellular adhesion molecule 1) and VCAM-1 (vascular cellular adhesion molecule 1), which allow macrophages to adhere to them and extrude from the circulation into the tissue [36]. Once the macrophage has extruded from the circulation, it phagocytoses the oxidized LDL (low-density lipoprotein) where it accumulates in the cytosol as droplets transforming the macrophage into a foam cell (the main component of the atherosclerotic plaque) [37]. Furthermore, macrophages are able to contribute to the ongoing local disruption by secretion of pro-inflammatory mediators such as TNFα and IL-6 as well as metalloproteinases, thereby enhancing vessel remodelling and plaque disruption [38].

In the vascular endothelium, similar to the adipose tissue, evidence implies that plasticity of macrophage polarization is crucial in influencing plaque outcome [39], as proven by the fact that atherosclerotic plaques that cause clinical events (stroke or myocardial infarction) are characterized by a high content of pro-inflammatory macrophages secreting a large amount of metalloproteinases [40]. A recent study by van Gils et al. [41] was able to once again underscore the importance of the macrophage in the development of atherosclerotic disease and the potential therapeutic benefits in preventing the accumulation of these cells. However, atherosclerosis is not only characterized by a profound change in macrophage numbers, but also by a striking change in polarization.

In comparison with early atherosclerotic plaque, in which macrophages express Arg1 (a marker of alternative activation), there is a switch to the expression of Arg2 (marker of classical activation) in the setting of mature plaques [42]. Indeed, hypercholesterolaemic apoe−/− mice fed a high-fat diet develop a growing amount of Ly-6Chi monocytes (capable of differentiating into M1 macrophages) in the peripheral blood, whereas the amount of Ly-6Clo monocytes remained similar throughout the same timespan. Furthermore, these Ly-6Chi monocytes developed a propensity to accumulate in atherosclerotic plaques where they are able to differentiate into M1 macrophages and thereby drive inflammation and plaque instability [43]. As we described above, M1 macrophages are able to secrete a wide amount of pro-inflammatory cytokines such as IL-6, IL-7 and the soluble ligand CD40, which is increased in patients with unstable angina and myocardial infarction, with higher levels of IL-6 predicting a worse outcome [44]. On top of this, M1 macrophages have also been implicated in smooth muscle cell proliferation [42] in vitro as well as the production of reactive oxygen species which not only are cytotoxic, but also have consequences for lipoprotein toxicity [45].

One of the most relevant and interesting discoveries regarding the potential effects of macrophage polarization over the last few years was made by Feig et al. [46]. His group was able to show that a reversal of hyperlipidaemia not only decreases the content of CD68+ cells (macrophages and foam cells), but also produces a switch from M1 to M2 macrophages in the atherosclerotic plaques of Reversa mice. In line with this, there was a reduction in the amount of inflammation seen in the plaque, as shown by reduced levels of MCP-1, TNFα and VCAM-1, but also an increase in the amount of collagen present in the plaque. This has proven to be of beneficial effect as it provides plaque stabilization through increased fibrosity [47]. Interestingly, although not leading to increased CD68+ efflux, the authors observed a further increase in the expression of M2 markers when the animals were treated with pioglitazone, a PPARγ activator supporting some of the data we have reviewed above that PPARγ indeed may support the conversion from M1 to M2 (Figure 4).

Figure 4 Model of the development of atherosclerosis

It is suggested that the state of inflammation created by the advent of obesity, hypertension, hyperlipidaemia and insulin resistance results in the extrusion and accumulation of macrophages and the formation of an early plaque inside the vessel wall. With continued inflammation there is increased recruitment of macrophages, which exude a phenotypic switch to an inflammatory phenotype (M1) thereby creating an ‘inflamed’ milieu, which further attracts immune cells thus creating a vicious circle. However, one could potentially, by influencing macrophage polarization, influence plaque stability through increased collagen deposition and a reduction in the release of inflammatory mediators.

Similar results were reported in humans by the group of Staels [48] who examined the content of carotid atherosclerotic plaques and was able to show that CD68+ cells that are MR (mannose receptor)-positive tend to be associated with areas of plaque that are cell rich and stable as well as expressing higher levels of IL-4 providing a cytokine environment for M2 polarization. On the other hand, CD68+ cells that were MR predominated in the lipid-rich necrotic core of the plaque. Additionally the group was able to show that there is a difference in the lipid content between the two macrophage groups [48]. The Cd68+MR cells were filled with more abundant larger lipid droplets than CD68+MR+ cells despite a lower capacity for cholesterol efflux. This means that these alternative macrophages are able to protect themselves against free cholesterol, which the authors suggested may be by increasing cholesterol esterification providing further support to the theory that alternatively activated macrophages are able to positively influence the atherosclerotic process [48].

Another theory that has been proposed as a possibility of explaining the advantage of M2 polarization is the ability of M2 macrophages to be more competent at clearing apoptotic cells and tissue debris, a process known as efferocytosis [49].

Finally, it has been recently observed that regulated accumulation of desmosterol underlies many of the homoeostatic responses, including activation of LXR (liver X receptor) target genes, inhibition of SREBP (sterol-regulatory-element-binding protein) target genes, selective reprogramming of fatty acid metabolism and suppression of inflammatory-response genes, observed in macrophage foam cells. These observations suggest that macrophage activation in atherosclerotic lesions results from extrinsic pro-inflammatory signals generated within the artery wall that suppress the homoeostatic and anti-inflammatory functions of desmosterol [50].


In obese animals [51] and humans [52], MCP-1 [CCL2 (CC chemokine ligand 2)] is up-regulated in the adipose tissue. This ligand of CCR2 (CC chemokine receptor 2) is necessary for monocyte and macrophage chemotaxis and is also implicated in regulating glucose uptake in insulin-sensitive cells. An animal line lacking this receptor was protected against obesity-induced inflammation and insulin resistance [53].

Another piece of the puzzle elucidating the role of the ATMs (adipose tissue macrophages) came when Patsouris et al. [54] showed that the ablation of M1 CD11c+ cells reduces insulin resistance and adipose tissue inflammation despite not affecting the level of adiposity.

Further evidence for the central role of classically activated macrophages in the development of insulin resistance and defective insulin signalling came from Hirosumi et al. [55]. They were able to show that ablation of Mapk8 [the gene encoding JNK (c-Jun N-terminal kinase) 1, a mediator of inflammation] results in increased sensitivity to insulin and enhanced Insr signalling in two different models of mouse obesity [55]. A similar mouse model was then used to clearly pinpoint adiposity-related inflammation rather than adiposity itself as the causative factor for insulin resistance. By selectively knocking out JNK1 from the haemopoietic department alone, Solinas et al. [56] was able to uncouple hepatic and adipose tissue inflammation from adiposity alone, strongly suggesting that the former is mainly due to macrophage activation and infiltration rather than adiposity alone.

In 2007 Lumeng et al. [23] showed that, in lean mice, the ATMs, rather than being silent and less numerous, are actually activated along the alternative pathway and express many genes seen in the M2 phenotype, such as Arg1, Mrc1 and Clec7a. However, in the setting of obesity there is a switch in the gene expression towards TNFα, IL-6 and IL-12 up-regulation characteristic of the classically activated macrophage (M1), suggesting that retaining the macrophage in its M2 polarized state may confer some protection towards metabolic disease [57].

The observation that M2 polarization might result from lipid-induced regulation through cholesterol metabolites such as desmosterol has clear implication also in the setting of insulin resistance. Since the nature of the factors that regulate the homing and differentiation/polarization of macrophages within the adipocytes is still elusive, we cannot exclude that cholesterol metabolites could be implicated. Indeed Spann et al. [50] recently suggested that a metabolite of the cholesterol pathway, desmosterol, may be responsible for inactivating gene transcription inside foam cells through potent induction of LXR (in itself an inflammatory response repressor). This suggestion opens up the possibility that, rather than cholesterol itself causing the acquisition of a pro-inflammatory phenotype in the context of atherosclerosis, it could be external stimuli such as inflammatory mediators and cellular debris causing this effect. Further research will be needed to determine whether the desmosterol pathway may be inhibited in atherosclerosis, thereby reducing the LXR activation and hence an increased inflammatory phenotype.


Although the role of insulin signalling and the part it plays in cell metabolism and growth have been widely studied, the role of insulin or the effects of insulin resistance on the macrophage have, so far, not been completely elucidated. Although most insulin signalling molecules are expressed in macrophages, there are some exceptions, the most noteworthy being IRS (insulin receptor substrate) 1 [58] and Glut-4 [59].

The Insr KO in myeloid cells

Brüning and co-workers described an unexpected reduction of atherosclerosis in ApoE mice with insulin resistance restricted to the haemopoietic compartment [60]. The potential explanation given by the authors was that insulin itself is able to induce a pro-inflammatory response in macrophages by increasing the secretion of pro-inflammatory markers such as IL-1β and IL-6 [60]. To further elucidate the effect of insulin on macrophages, the group then went on to examine the effect of a high-fat diet on systemic inflammation and obesity in the same mouse model. Surprisingly, the authors found that the IRmyel−/− mice were protected from diet-induced adipose tissue inflammation, despite similar weight gain and body fat percentage increase [61]. It appears that knocking down the Insr prevented macrophage recruitment into adipose tissue, suggesting that insulin signalling is critical for macrophage recruitment into white adipose tissue. The authors went on to suggest that this phenotype could be due to a down-regulation of MMP (matrix metalloproteinase) 9, a protease previously proven to be directly involved in tissue invasion of macrophages [62]. Overall, the authors came to the conclusion that, despite the positive effect insulin (and thereby insulin signalling) has on target tissues such as the liver, white adipose tissue and skeletal muscle, its action on macrophages may be deleterious through its ability to promote a pro-inflammatory state. However, data from human monocytes from insulin-resistant subjects described an anti-inflammatory action of insulin based mostly on the inhibition of the TLR (toll-like receptor)-4/NF-κB (nuclear factor κB) pathway [6365].

A potential explanation for this discrepancy between genetics models and human physiology may be linked to repression of the ability of FoxO (Forkhead box O) 1 to activate the TLR-4 promoter [66], a pathway we will examine separately below.

LDLRKO (LDL receptor knockout) with bone marrow transplant from Insr-knockout

As mentioned above, the effects of insulin resistance on macrophages remain controversial. In 2006 Tall and co-workers transplanted macrophages from transgenically rescued Insr-knockout mice (Insr−/−) into the pro-atherogenic LDL−/− mouse model. When fed a western diet, these animals developed larger and more complex atherosclerotic lesions with increased necrotic cores when compared with bone marrow transplant from Insr+/+ mice, whereas cholesterol and TAG levels were similar between the two groups [67]. These insulin-resistant macrophages showed a decreased ability to withstand endoplasmic reticulum stress and an enhanced susceptibility to stress-induced apoptosis, which the authors concluded was due to a reduction in Akt signalling, which is in a way contradictory to what Brüning and co-workers [60] had demonstrated. The reasons for the differing results between the two studies so far are not completely clear. Some may pertain to different areas of the aorta being analysed. Furthermore, the diet used to induce atherosclerosis differed in the two groups. Brüning and co-workers used a diet containing 5% cholesterol [60]; Tall and co-workers used a diet containing only 0.2% cholesterol [67].

The AKT and FoxO axis

Protein kinase B (also known as AKT), a central mediator regulating multiple cellular functions, is one of the main downstream targets of insulin's intracellular actions. Previous studies have shown that the AKT pathway is impaired in insulin resistance, causing a loss of this atheroprotective pathway. Three similar AKT isoforms exist, but, despite their homology, they appear to carry out non-redundant functions [68]. Although whole-body Akt1-knockout mice show general growth impairment [69], purely endothelial Akt1-deficient mice show reduced endothelial NO synthethase phosphorylation, NO release and endothelial cell migration as well as impaired VEGF (vascular endothelial growth factor) and ischaemia-induced angiogenesis [70]. Akt2-knockout mice, in contrast, show impaired glucose tolerance and insulin resistance, whereas Akt3-knockout mice show a selective defect in brain development [71]. The effect of AKT1 and AKT3 has been subject to some debate, but has been investigated. Fernandez-Hernando et al. [72] investigated the effect of loss of Akt1 in the pro-atherogenic ApoE mouse model. The authors found that the loss of Akt1 resulted in increased general athero- and arteriolo-sclerosis despite similar lipid profiles. The authors found that the lack of Akt1 leads to increased endothelial cell apoptosis and reduced eNOS (endothelial nitric oxide synthase) phosphorylation, which could be rescued by re-introduction of Akt1 through an adenoviral vector. Surprisingly, despite an increased infiltration of pro-inflammatory macrophages into the intima, the macrophages themselves were shown to take up less LDL (which would favour an anti-atherosclerotic effect). Furthermore, bone marrow transplantation experiments did not show any evidence that macrophage loss of Akt1 is able to account for this phenotype. Therefore the authors speculated that this effect is strongly offset by the reduced availability of antioxidants causing endothelial cell damage.

Ding et al. [73] then went on to examine the effect of Akt3 deficiency on the development of atherosclerosis. Similar to the Akt1-knockout model, the loss of Akt3 leads to more severe atherosclerosis as well as increased macrophage infiltration into atherosclerotic lesions. In this case, however, the authors were also able to show through bone marrow transplantation that it is the loss of Akt-3 in the haemopoietic compartment that leads to this phenotype [73]. Despite Akt3 only being a minor isoform in macrophages, the ablation of it leads to an intracellular ablation of cholesterol esters and foam cell formation through an increased lipoprotein uptake as well as an increased ACAT-1 (acyl-CoA:cholesterol acyltransferase 1) activity.

One of the substrates of AKT are the Forkhead box transcription factors (FoxO1, FoxO3a and FoxO4a). In response to AKT-mediated phosphorylation, FoxOs are excluded from the nucleus and thus deactivated. They are able to repress eNOS activity in endothelial cells [74] and thus are a putative mediator of increased atherosclerosis in diabetics (where AKT signalling is reduced).

Indeed, Accili and co-workers showed that knocking out FoxO1, FoxO3a and FoxO4a in endothelial cells leads to a decrease in atherosclerosis in the LDLRKO model by increasing NO production and reducing oxidative stress and inflammation [75].

Interestingly, however, the same group also found that conditionally knocking out FoxOs in the myeloid compartment alone leads to the generation of more severe atherosclerosis through an expansion of the myeloid production of neutrophils and monocytes (the initiators of atherosclerosis) as well as an up-regulation of iNOS (inducible nitric oxide synthase), thus leading to increased oxidative stress in macrophages [76].

The somewhat controversial and opposing data that have so far been discovered about FoxO suggest different actions in different tissues and much more will have to be done in order to define what the exact impact of FoxO signalling is on the development of atherosclerosis.

PPARs and their effects on macrophage polarization

PPARs function as the body's sensor for fatty acids and therefore are able to orchestrate glucose and lipid metabolism depending on the source of energy present. Odegaard et al. [29] showed the intimate involvement of PPARγ and PPARδ in the ability of macrophages to polarize towards the M2 state and thereby potentially reduce the sequelae of obesity.

PPARγ is induced by IL-4 and is able to up-regulate β-oxidation and mitochondrial biogenesis in macrophages and therefore it is predisposed towards the alternative pathway. Indeed, when mice lacking PPARγ in their macrophages are challenged with a high-fat diet, there is a specific down-regulation in alternatively activated macrophages which in turn leads to increased local inflammation in the adipose tissue. Furthermore, these animals exhibited significantly higher glucose intolerance, weight gain and reduced insulin sensitivity when compared with wild-type animals [29].

A similar propensity to develop obesity and insulin resistance through a lack of M2 polarization is seen when constitutionally knocking out PPARδ in the haemopoietic compartment [77]. Despite it not being necessary to up-regulate oxidative metabolism, PPARδ, upon induction by IL-4, synergizes with STAT6 to control the expression of several genes required for alternative activation and M2 polarization, such as MGl1, Mgl2 and MRC2, while at the same time down-regulating genes responsible for classical activation including MCP-1, TNFα and IL-6. In the liver the Kupffer cell represents the local macrophage and its activation has been implicated in both obesity-induced fatty liver disease and insulin resistance [78]. Similarly to ATMs, Kupffer cells display plasticity in their activation, allowing them to develop into similar pro- and anti-inflammatorily orientated cells. Since liver insulin resistance alone is able to create an organism-wide insulin resistance, interest developed into examining whether shifting Kupffer cells towards alternative activation could have an impact on the development of local inflammation and insulin resistance. Indeed Kupffer cells in which PPARδ was knocked out showed a diminished capacity to undergo alternative activation, leading to a dramatic reduction in insulin sensitivity [79].

To summarize, one can say that through their abilities to up-regulate oxidative capabilities (PPARγ) and specific genes (PPARδ), these nuclear receptors are able to influence the ability of macrophages to be alternatively activated and thereby reducing local inflammation and in turn increase glucose tolerance and insulin sensitivity.


TLR-4 is a member of the TLR family, which play a role in pathogen recognition and activation of the innate immune system. TLR-4 has been recognized as being able to activate NF-κB in response to short-chain fatty acids, eliciting a further link that nutrient excess itself is able to activate the immune system [80]. Several lines of investigation have linked TLR-4 to M1 polarization and increased adipose tissue inflammation. First, TLR-4 has been shown to be highly expressed by M1 polarized macrophages, whereas it is down-regulated in M2 macrophages [81]. Secondly, TLR-4 deficiency has been shown to be protective against obesity-induced insulin resistance and adipose tissue inflammation [82]. However, it was only proven recently by Hasty and co-workers that TLR-4 receptor deficiency promotes alternative activation of macrophages. In line with this, the authors were able to produce a modest reduction in adipose tissue inflammation in the haemopoietic and the general TLR-4−/− model [83]. Interestingly, despite the increase in alternatively activated macrophages in the adipose tissue and the reduction in adipose tissue inflammation, the authors did not find any difference in systemic insulin resistance, suggesting that the shift towards alternate activation was either not pronounced enough or that other mechanisms contribute to insulin resistance in this model.


Members of the tribble family of pseudokinases have only very recently been implicated in the differentiation process of M2 macrophages. Satoh et al. [84] showed that mice lacking Trib1 in the haemopoietic compartment developed hypertriglyceridaemia and insulin resistance. The authors went on to demonstrate that the lack of Trib1 causes an inability of the bone marrow to develop M2-like macrophages, which could then be activated in adipose tissue and the liver. Of further note in that single study on the effect of Trib1 on the development of insulin resistance and obesity is the fact that the authors found smaller adipose tissue mass in the Trib1 animals. Upon supplementing M2 macrophages, the adipose tissue mass increased when compared with animals injected only with PBS. The authors conclude that in the absence of Trib1 the reduction in M2 macrophages leads to a lipodystrophic phenotype which has been previously implicated with metabolic disorders [85]. The idea that a reduction in M2 macrophages may be involved in the effect of lipodystrophy, which as discussed above can have big implications on the development of insulin resistance, opens new doors for further research into this field.

ADAM17 (a disintegrin and metalloproteinase 17) and TIMP (tissue inhibitor of metalloproteinases)-3

TIMPs are endogenous inhibitors of MMPs which can act on a variety of substrates localized in the plasma membrane to generate inflammation, growth, migration and metabolic signals [86]. Unrestricted MMP activity has been linked to increased extracellular membrane destruction in the arterial wall, thus predisposing an arterial plaque to rupture [87].

One of the TIMPs we will consider in the present review is TIMP3, which we have identified as a gene modifier for diabetes and vascular inflammation in Insr+/− mice [88]. In contrast with other TIMPs, TIMP3 also preserves the ability to inhibit the TNFα-converting enzyme also named ADAM17. ADAM17 is a sheddase for several growth factors, receptors and adhesion molecules [89], including factors such as TNFα, soluble ICAM-1 and soluble IL-6 receptor, all known to be elevated in the blood of patients with diabetes and its vascular complications [89]. Loss of TIMP3 on Insr heterozygosity was initially associated with vascular inflammation and development of small atherosclerotic plaques [90]. ADAM17 inhibition either by genetic approach or by chemical inhibitor partially rescued the pro-inflammatory and pro-diabetogenic effect of low TIMP3 expression. The relevance of TIMP3 as a regulator of signals relevant for diabetes and atherosclerosis was supported by the observation of low TIMP3 expression with elevated ADAM17 activity in the atherosclerotic plaque, and in the skeletal muscle and foot ulcers of diabetic patients [91].

TIMP3 is secreted by cells into the extracellular matrix where it is bound to glycosaminoglycans. Whether it is the TIMP3 released from stromal or inflammatory cells that tackles inflammation and preserves glucose metabolism is still unclear. However, a few observations point to monocytes as a major source for TIMP3 in the context of metabolic and vascular injuries. In fact, Johnson et al. [92] showed that during the conversion of non-foamy macrophages to foam cells there is a down-regulation of TIMP3 and low TIMP3-expressing macrophages maintain a high invasive capacity in vascular lesions. We also found that monocytes from insulin-resistant subjects carry a low expression of TIMP3 associated with low IRS2 expression [93]. Low Timp3 expression in monocytes was also found to be associated with increased intimal medial thickness, suggesting that restraining the metalloproteases, inhibited by TIMP3 such as MMP9 and ADAM17, might protect from progression of intimal hyperplasia, one step of atherosclerotic plaque progression.

One could speculate that this down-regulation of TIMP3 during the classical activation and transformation of a monocyte to an M1 macrophage leads to increased MMP activity and thus increased destruction of the extracellular matrix. Rescuing this down-regulation of TIMP3 may provide a means to reduce the effect of M1 activation on plaque stability. To answer this question we engineered a mouse model to overexpress TIMP3 under the control of the CD68 promoter. CD68 is a marker for classical macrophage differentiation, allowing us to test the hypothesis that forced release of TIMP3 surrounding the inflammation could provide a brake to signals detrimental for atherosclerosis and diabetes occurrence [94,95]. Studies performed with models of diabetes or atherosclerosis provide compelling confirmation to our hypothesis. In particular, when we used adoptive transfer to monitor the homing of high TIMP3-expressing monocytes towards the atherosclerotic plaques, we clearly observed that they were blocked at the endothelial layer with reduced penetration into the inflamed arterial wall [94].

More recently it has been shown that macrophage ADAM17 deficiency augments CD36-dependent apoptotic cell uptake and the linked anti-inflammatory phenotype. CD36 is one of the main regulators for apoptosis and efferocytosis in macrophages, and blocking ADAM17-induced shedding of CD-36 from the surface of macrophages was shown to increase the uptake of apoptotic cell debris [96]. Enhancing efficient phagocytosis in macrophages leads them to secrete anti-inflammatory cytokines such as IL-10 and thereby actively dampen inflammation.


Insulin resistance and diabetes increase the incidence of atherosclerosis in the coronary [97], extracranial [98] and peripheral [99] arteries. This effect is probably due to several causes, including hyperglycaemia, insulin resistance, endothelial cell dysfunction and excess FFAs. In the present review, however, we have decided to concentrate on the step that is common to both diseases, which is the continuous low-grade increase in inflammation that is seen in the adipose tissue as well as the vasculature. In both tissues there is not only an increase in inflammatory markers, but also a change in the polarization of macrophages, the local resident inflammatory co-ordinator.

Although the present review has focused on the importance of macrophage polarization in the setting of diabetes, atherosclerosis and inflammation, we do acknowledge that many other immune cells, including dendritic cells, B-cells and T-cells, are involved in the process of atherosclerosis as well as diabetes. However, the size of this field precludes discussion of all of the mediators involved [100] in the present review. We would like to shortly turn our attention to T-cells, where we believe there are still many open questions and interesting recent developments. Indeed, up to 50% of cells present in a mature atherosclerotic plaque are T-cells with CD4+ (T-helper) cells being predominant over CD8+ (T-cytotoxic) cells [101]. Naïve T-cells are rarely found in atherosclerotic plaques as their activation usually occurs in the secondary lymphoid organs upon presentation of an antigen. CD4+ cells are activated in response to MHC II, whereas CD8+ cells are activated upon presentation of MHC I, then migrate to the affected tissue where the presentation of the same antigen by an APC (antigen-presenting cell) results in the definite activation of the T-cell.

However, recent publications offer interesting new views on the modulation of CD8+ metabolism to potentially influence the development of atherosclerosis. Quiescent CD8+ cells are believed, like most other tissues, to use oxidative phosphorylation to interchangeably break down glucose, fatty acids and amino acids to provide for their energy needs [102]. However, Warburg [103] described in 1956 that proliferating T-cells, despite enough available oxygen to support oxidative phosphorylation, switch to glycolysis to provide energy. Interestingly, in contrast with the proliferating cell group, CD8+ memory cells revert again to oxidative phosphorylation as an energy source. In addition to this, CD8+ memory cells, not naïve or proliferating cells, maintain substantial SRC (spare respiratory capacity), which is available to the cell to produce energy under conditions of increased work or stress [104].

From these observations, one could speculate that, since T-cells are involved in the setting of atherosclerosis, in which they expand rapidly thus potentially switching to glycolysis but maintaining a large SRC, finding a way to revert them back to oxidative phosphorylation could potentially reduce the atherosclerotic burden by using up fatty acids inside the plaque.


Insulin resistance and atherosclerosis are two intimately linked disorders thought for decades to be primarily metabolic in their nature. With the availability of new experimental models and diagnostic techniques in patients we are now facing a change in our perspective of the two disorders. Whether insulin resistance and atherosclerosis will be clustered in one entity called ‘immunometabolic defects’ is under current discussion since it may change our preventive and therapeutic policies in the long term. Despite the evidence that both disorders have aspects of the so-called sterile inflammation, the nature of the auto-antigen remains elusive. Only the identification of the triggering auto-inflammatory factors(s) will definitively prove the immunometabolic basis of insulin resistance and atherosclerosis.


This manuscript was funded in part by Progetti di Ricerca di Interesse Nazionale (PRIN) 2009, Florinash FP7, European Consortium of the Early Treatment of Diabetic Retinopathy (EuRhythDia) FP7, European Foundation for the Study of Diabetes (EFSD)/Lilly 2012 and Associazione Italiana per la Ricerca sul Cancro (AIRC) 2013 (all to M.F.). Robert Stöhr is supported by a fellowship from the Deutsche Herzstiftung and the University of Rome Tor Vergata.

Abbreviations: ADAM17, a disintegrin and metalloproteinase 17; ApoE, apolipoprotein E; Arg, arginase; ATM, adipose tissue macrophage; eNOS, endothelial nitric oxide synthase; FFA, non-esterified (‘free’) fatty acid; FoxO, Forkhead box O; ICAM-1, intercellular adhesion molecule 1; IL, interleukin; IFNγ, interferon γ; Insr, insulin receptor; IRS, insulin receptor substrate; JNK, c-Jun N-terminal kinase; LDL, low-density lipoprotein; LDLRKO, LDL receptor knockout; LXR, liver X receptor; MCP-1, monocyte chemoattractant protein 1; MMP, matrix metalloproteinase; MR, mannose receptor; NF-κB, nuclear factor κB; PGC1β, peroxisome-proliferator-activated receptor γ co-activator 1β; PPAR, peroxisome-proliferator-activated receptor; SRC, spare respiratory capacity; STAT, signal transduction and activator of transcription; T2DM, Type 2 diabetes mellitus; TAG, triacylglycerol; TIMP, tissue inhibitor of metalloproteinases; TLR, toll-like receptor; TNFα, tumour necrosis factor α; VCAM-1, vascular cellular adhesion molecule 1


View Abstract