Insulin Resistance in Macrophages Alters Their Metabolism and Promotes an M2-Like Phenotype

Obesity and type 2 diabetes (T2D) are characterized by chronic low-grade inflammation. Macrophages, central mediators of inflammation, are causally linked to the development of obesity-related pathological conditions. During obesity and T2D, the insulin signaling pathway is impaired, leading to hyperinsulinemia and insulin resistance (1).

It is increasingly being recognized that metabolic pathways play an important role in determining immune cell function (2, 3). Although macrophages contribute to insulin resistance–associated inflammation, it is not known how insulin resistance affects macrophage metabolism and function. Macrophages express all components of insulin signaling cascade, implying a responsive insulin pathway. Insulin mediates its signals through insulin receptor (IR) and insulin-like growth factor 1 receptor (IGF1R). Recent evidence demonstrated that IGF1R is suppressed in M1-type–activated macrophages and upregulated upon IL-4 stimulation (M2 differentiation), contributing to macrophage activation phenotype (4). Genetic ablation of IR in macrophages results in an anti-inflammatory phenotype (5). In contrast, macrophages that lack IGF1R acquire a more inflammatory phenotype in response to proinflammatory stimuli (4, 6). However, IRS2, which mediates both insulin and IGF1 signaling, suppresses alternative activation of macrophages (7). In addition, glucose, insulin, and palmitate treatment of human monocyte–derived macrophages results in a metabolically activated phenotype, a distinct M1 phenotype that is characteristic for proinflammatory ATMs found in obesity (8).

Insulin signaling triggers the activation of the PI3K/Akt pathway and the initiation of a signaling cascade of downstream molecules regulating cell metabolism. Genetic ablation of Akt kinase isoforms differentially affects macrophage activation phenotype. Deletion of Akt2 renders macrophages M2-like, whereas ablation of Akt1 isoform renders them prone to M1 differentiation (9).

Another important mediator of insulin signaling and macrophage activation and metabolism is mTOR (3, 10, 11). mTOR is implicated in the propagation of insulin signaling, acting either upstream or downstream of Akt. Thus, mTORC2 phosphorylates and activates Akt at Ser⁴⁷³, whereas mTORC1 is activated by Akt through the phosphorylation and inhibition of TSC1/TSC2 complex. Macrophage specific deletion of Raptor, an mTORC1 component, elevates M2 macrophage population in LysM^CreRptor^fl/fl mice (12), consistent with the effect of TSC1 deletion that leads to hyperactivation of mTORC1 and decrease of M2 macrophage population (13, 14). In contrast, lack of TSC2 in macrophages enhanced the expression of M2 polarization markers in macrophages (15), suggesting that mTOR signaling regulates macrophage activation phenotype. mTOR is important for the regulation of several metabolic pathways, including glycolysis, pentose phosphate pathway, and lipid biosynthesis via phosphorylation of 4E-binding protein 1 (4E-BP1) and S6 kinase (16, 17).

The metabolic status of macrophages directs their activation status. Inflammatory macrophages are characterized by increased rates of glycolysis (18, 19). Deletion of glucose transporter 1 (Glut1), essential for glucose uptake in macrophages, results in reduced production of proinflammatory cytokines (20). Alternatively activated macrophages, in contrast, display increased glutamine uptake and catabolism and oxidative phosphorylation (21, 22), but they also uptake more glucose compared with naive cells (23). Tumor-associated macrophages (TAMs) acquire an M2-like phenotype characterized by increased glycolysis (24) and the end product of glycolysis, lactic acid, produced in the tumor microenvironment triggers the expression of M2 activation markers (25), whereas IL-4–induced M2 macrophages display increased glycolysis and increased oxidative phosphorylation (26), highlighting the complexity of metabolic adaptation of macrophages under different conditions (19).

In this study, we examined the impact of insulin resistance on macrophage activation and polarization phenotype as well as their metabolism. First, we confirmed that macrophages develop insulin resistance in the context of systemic insulin resistance. We found that insulin-resistant macrophages had defective Akt2 and sustained mTORC1 activation, increased glycolysis, and possessed an M2-like phenotype, which we refer to as M-insulin resistant (M-InsR). Mice harboring M-InsR macrophages exhibited reduced acute systemic inflammatory response and reduced lung injury in polymicrobial sepsis but also demonstrate reduced bactericidal capacity. Thus, we describe a novel M2-like phenotype induced by insulin resistance and associated with obesity-related changes in acute inflammatory responses.

C57BL/6 (wild-type [WT]), Akt2^−/−, Akt1^−/−, LysM^CreIgf1R^fl/fl, and Igf1R^fl/fl mice were kept in a pathogen-free animal facility of the Institute of Molecular Biology and Biotechnology in Heraklion, Crete. For the diet-induced obese group and diet-induced, insulin-resistant group, 8-wk-old male C57BL/6 mice were fed a high-fat diet (HFD; 60% energy from fats; catalog number PF4051/D) or a normal chow diet (catalog number 4RF21; purchased from Mucedola) for 7 d or 10 wk, according to the protocol (short-term HFD or long-term HFD). All procedures were conducted in compliance with protocols approved by the Animal Care Committee of the University of Crete, School of Medicine (Heraklion, Crete, Greece) and from the Veterinary Department of the Region of Crete (Heraklion, Crete, Greece), license number 150760-07/20/17.

Primary thioglycolate-elicited peritoneal macrophages (TEPMs) and alveolar macrophages were cultured in complete medium (DMEM; Life Technologies, Carlsbad, CA) supplemented with 10% (v/v) FBS, 10 mM l-glutamine, 100 IU/ml penicillin, and 100 mg/ml streptomycin. Escherichia coli–derived LPS (100 ng/ml; Sigma-Aldrich, St. Louis, MO), insulin (100 nM; Humulin R, Eli Lilly and Company), rapamycin (20 nM; Sigma-Aldrich), 2-deoxy-d-glucose (2DG; 2.5 mM; Sigma-Aldrich), and FCCP (Sigma-Aldrich) were used when indicated.

Cells were cultured in a 96-well plate in a density of 1 × 10⁵ cells per well. Uptake of glucose was measured in a beta counter. Alternatively, glucose uptake was measured with a kit from Abcam (ab136955) according to the manufacturer’s instructions.

The cecal ligation and puncture (CLP) model is a widely used model of polymicrobial sepsis (27) that can indirectly lead to acute lung injury. Mice were anesthetized with ketamine (80–100 mg/kg) and xylazine (5–15 mg/kg) injected i.p. A midline incision (1.5–2 cm) was made to reach the cecum and exteriorize it. The distal 0.5-cm portion of the cecum was ligated with a 4-0 silk ligature suture without interrupting intestinal continuity. The cecum was punctured twice with a 16-gauge needle to become perforated and squeezed softly. The cecum was returned to the abdomen, and the incision was closed in layers with a 2-0 silk suture. After the procedure, animals were fluid-resuscitated with sterile saline (1 ml) injected s.c. Sham controls were subjected to the same procedures as CLP without ligation and puncture of the cecum.

Bronchoalveolar lavage fluid (BALF) was collected by intratracheal infusion of normal saline (30 ml/kg) to the lung. The supernatant was used for measurements of cytokine levels. Alveolar cells were obtained by performing four more washes of the alveolar space with 1 ml of normal saline and stained for flow cytometry. Blood serum was collected from the right ventricle of the heart using a 1-ml syringe. Samples were centrifuged at 5000 rpm for 10 min at 4°C. Serum was collected and used for cytokine levels measurements.

Cytokine concentration for TNF-α, IL-6, and MIP-2a was determined by ELISA at the indicated time points using ELISA kits (R&D Systems) according to the manufacturer’s instructions.

For histology purposes, lungs were perfused with PBS through the right ventricle. An incision at the left atrium allowed outflow of the blood. Lungs were inflated intratracheally with 10% formalin at 25 cm H₂O pressure, fixed overnight at 4°C, and stored in 70% ethanol before embedding in paraffin. Lung tissue sections of 5 μm were prepared and further deparaffinized and rehydrated. Sections were stained with H&E and evaluated by a pathologist blinded to the interventions. To perform histological assessment of lung injury, five independent variables were evaluated (neutrophils in alveolar spaces, neutrophils in interstitial spaces, hyaline membranes, proteinaceous debris filling the airspaces, and alveolar septal thickening) and weighted according to the relevance ascribed to by the Official American Thoracic Society Workshop Report on Features and Measurements of Experimental Acute Lung Injury in Animals (28). The resulting injury score is a continuous value between 0 and 1.

Extracellular acidification was measured using glycolysis assay kit (Abcam, Cambridge, U.K.) according to the manufacturer’s instructions. Briefly, TEPMs were seeded at a density of 1 × 10⁵ cells per well and incubated overnight in CO₂ incubator. Then, CO₂ was purged, and medium was replaced with respiration buffer containing glycolysis assay reagent along with different stimulatory compounds. Measurement was performed using time-resolved fluorescence in a PerkinElmer VICTOR Series ×4 microplate reader.

RNA from TEPMs or total lung was isolated using TRIzol Reagent (Life Technologies). One microgram of total RNA was used for cDNA synthesis (TAKARA, Shiga, Japan). The SYBR Green method was followed in the PCR. Ribosomal protein S9 served as the housekeeping gene. Annealing was carried out at 60°C for 30 s, extension was at 72°C for 30 s, and denaturation at 95°C for 15 s for 40 cycles in a 7500 Fast Real-Time PCR System (Life Technologies/Applied Biosystems). The amplification efficiencies were the same as the one of ribosomal protein S9 as indicated by standard curves of amplification, allowing us to use the following formula: fold difference = 2^−(ΔCtA − ΔCtB), in which Ct is the cycle threshold.

For measurement of myeloperoxidase (MPO) determination, 50-mg/ml tissues were homogenized in 50 mmol/l phosphate buffer (pH 6.0) with 0.5% hexadecyltrimethylammonium bromide using mortar and pestle. Samples were frozen and thawed three times and centrifuged for 10 min at 10,000 × g, and then supernatants were stored at −20°C until assay. MPO was determined in 96-well plates using a modification of the method described (27). Briefly, 10-μl sample was added to 190 μl assay buffer (phosphate buffer 50 mM [pH 6.0] containing 0.167 mg/ml o-dianisidine and 0.0005% H₂O₂; Sigma-Aldrich). Absorbance at 450 nm was measured in a microtiter reader at 15 min.

Expression of arginase 1 protein levels was determined by flow cytometry cell surface and intracellular staining, as previously described (29). Cells isolated from BALF were incubated with PerCP/Cy5.5 anti-mouse CD11c (BioLegend, San Diego, CA). Also, cells isolated from peritoneal cavity and from lung after homogenization were incubated with PE anti-mouse F4/80 (BioLegend) for cell surface staining; fixation and permeabilization (BD Fixation and Permeabilization Solution Kit; BD Biosciences) were performed for intracellular staining with mouse monoclonal anti-mouse arginase 1 (BD Biosciences). Allophycocyanin rat anti-mouse IgG1 (BioLegend) was used as secondary Ab for arginase 1 staining. The proper isotype controls were used in each case. The flow cytometry events were acquired in a MoFlo Legacy Cell Sorter (Beckman Coulter) and analyzed with Summit Software. Flow cytometry events were gated first based on forward and side scatter and then CD11c⁺ cells (alveolar macrophages) or F4/80 cells (peritoneal, lung interstitial macrophages) were selected to evaluate expression of arginase 1. TEPMs were incubated with MitoTracker Red CM-H₂X ROS (Molecular Probes, Invitrogen) in cell culture at a final concentration of 400 nM (Life Technologies) for 20 min at 37°C; the cells were collected and analyzed with flow cytometry for mitochondrial reactive oxygen species (ROS) determination.

Protein lysates from macrophages were resuspended in radioimmunoprecipitation assay buffer containing phosphatase and protease inhibitors (complete; Roche, Basel, Switzerland). Protein concentration of samples was determined using bicinchoninic acid kit. Twenty micrograms of protein was electrophoresed on 10% denaturing polyacrylamide gel prior to wet transfer to 0.45-μm nitrocellulose membrane (Macherey–Nagel, Germany). Briefly, after blocking with 5% BSA PBS (pH 7.4) for an hour at room temperature, the membranes were incubated with rabbit polyclonal anti-mouse p-Akt (Ser473) Ab (Cell Signaling Technology), rabbit polyclonal anti-mouse Akt Ab (Cell Signaling Technology), rabbit polyclonal anti-mouse p-Akt2 (Ser474) Ab (Cell Signaling Technology), rabbit polyclonal anti-mouse Akt2 Ab (Cell Signaling Technology), rabbit polyclonal anti-mouse p-S6 (235/236) Ab (Cell Signaling Technology), rabbit polyclonal anti-mouse p–4E-BP1(Thr37/46) Ab (Cell Signaling Technology), rabbit polyclonal anti-mouse inducible NO synthase (iNOS) Ab (Abcam), goat polyclonal anti-mouse arginase 1 (Abcam), goat polyclonal anti-mouse Fizz1 Ab (Abcam), rabbit polyclonal anti-mouse IGF-1Rβ Ab (Cell Signaling Technology), rabbit monoclonal anti-mouse IR-β (Cell Signaling Technology), or mouse monoclonal anti-mouse β-actin (Abcam) at 4°C overnight. The membranes were then incubated with 40 ng/ml peroxidase-conjugated anti-rabbit, anti-mouse, or anti-goat secondary Ab (Santa Cruz Biotechnology), respectively, for 1 h at room temperature, followed by reaction with Chemiluminescent HRP Substrate (LumiSensor; GenScript).

For insulin tolerance test, mice were fasted overnight and then injected i.p. with insulin (0.75 IU/kg). Glucose measurements were received at the following time points: 0, 15, 30, 60, and 120 min. For pyruvate tolerance test, mice were fasted overnight and then injected i.p. with 2 g/kg pyruvate. Glucose measurements were received at the following time points: 0, 30, 60, and 120 min. For glucose tolerance test, mice were fasted overnight and then injected i.p. with glucose (1 mg/g). Glucose measurements were received at the following time points: 0, 30, 60, and 120 min.

Phagocytic capacity of macrophages was measured using Vybrant Phagocytosis Assay Kit (Molecular Probes) according to the manufacturer’s instructions. Briefly, TEPMs were cultured in 96-well plates and incubated with heat-inactivated, fluorescein-labeled E. coli (K-12 strain) BioParticles for 2 h. Extracellular fluorescence was quenched by trypan blue, and phagocytic capacity was quantified by measuring fluorescence intensity of the uptaken particles using a PerkinElmer VICTOR Series ×4 microplate reader.

The results were analyzed using one-way or two-way ANOVA with Bonferroni multiple comparison posttest. Comparison of nonparametric results between different groups was performed by Mann–Whitney U test using GraphPad InStat software (GraphPad, San Diego, CA). The p values <0.05 were considered significant. Results are expressed as mean ± SD or as median (5th to 95th percentiles), as indicated, and are representative of at least three independent experiments.

Monocytes and macrophages are responsive to insulin signals through IR, resulting in phosphorylation of Akt kinase (30, 31). To determine whether macrophages develop insulin resistance, TEPMs from mice fed with HFD for a short term, being in a hyperinsulinemic environment in vivo, were isolated. These mice exhibited insulin resistance and glucose intolerance although the body weight has only mildly increased (Supplemental Fig. 1A–D). TEPMs from HFD-fed mice showed reduced Akt phosphorylation following 30-min insulin stimulation (Fig. 1A), and expression levels of IR-β were significantly lower (Fig. 1B) compared with TEPMs isolated from normal diet–fed counterparts. Basal Akt phosphorylation did not differ in TEPMs from HFD macrophages although IR was reduced, potentially because it was activated by IR-independent signals. TEPMs have the advantage to retain the phenotype obtained in vivo, in contrast to in vitro–differentiated bone marrow macrophages. Furthermore, TEPMs from WT mice were exposed for 48 h to high insulin concentration (100 nM) in an in vitro model of insulin resistance (32). Long-term high insulin exposure also resulted in reduction of Akt phosphorylation (Supplemental Fig. 2A) following a 30-min insulin restimulation and IR expression levels (Supplemental Fig. 2B).

Insulin resistance results in deregulation of the Akt/mTORC1 pathway. Akt2 is the major Akt isoform mediating insulin signals involved in glucose homeostasis (33, 34), and mice lacking Akt2 spontaneously develop insulin resistance and diabetes mellitus-like syndrome (35, 36). In line with these reported findings, TEPMs isolated from Akt2^−/− mice displayed dramatically reduced Akt phosphorylation in response to insulin stimulation (Fig. 1C) and exhibited insulin resistance (Supplemental Fig. 1A–D). Reduced Akt2 phosphorylation was also observed in TEPMs from HFD-induced, insulin-resistant mice (Fig. 1D). Akt1 is phosphorylated in response to insulin to propagate growth pathways (37). In insulin-resistant macrophages, although Akt1 was activated at basal levels, its phosphorylation was not further induced after insulin stimulation (Fig. 1E).

Phosphorylation of Akt kinases leads to activation of mTORC1. Hyperactivation of mTORC1 is linked to insulin resistance, being a key regulator of insulin sensitivity (38). Indeed, macrophages from HFD-fed mice (Fig. 2A) or Akt2^−/− macrophages (Fig. 2B) had elevated basal mTORC1 activation, as demonstrated by increased phosphorylation of mTORC1 targets, S6 (p-S6) and 4E-BP1 (p–4E-BP1). In contrast, the absence of Akt1 in macrophages led to reduced mTORC1 phosphorylation and activation upon insulin stimulation (Supplemental Fig. 2C).

Insulin signals via both IR and IGF1R. Insulin resistance in macrophages results in suppression of IGF1R (Supplemental Fig. 2D). Expression of both IR and IGF1R in Akt2^−/− macrophages was not affected, indicating that insulin resistance occurred at the level of Akt2 signaling (Supplemental Fig. 2E). To determine whether IR, IGF1R, or both receptors participate in insulin signaling in macrophages, we treated TEPMs with different doses of insulin and measured Akt2 phosphorylation and IR levels. The results showed that reduction of p-Akt2 or IR expression occurred at 1 nM, concentration that mediates signals via IR (39), and gradually decreased up to 100 nM, concentration that also mediates signals via IGF1R (Supplemental Fig. 2F). Moreover, IGF1 stimulation at 2, 6 nM, which mediates its action via IGF1R, and not IR (39), activated Akt2 phosphorylation in WT macrophages, but not in macrophages derived from HFD-fed mice, suggesting that both IR and IGF1R participate in insulin resistance in macrophages (Fig. 2F). Indeed, both IR and IGF1R were reduced in macrophages treated with high levels of insulin (Supplemental Fig. 2B) or from HFD-fed mice. IGF1R signals also activate the Akt/mTORC1 pathway, and recent evidence suggested that IGF1 through IGF1R regulates macrophage activation (4). To determine if IGF1R links insulin to Akt2/mTORC1, we generated LysM^CreIgf1R^fl/fl mice lacking IGF1R selectively from macrophages, which were not systemically insulin resistant (Supplemental Fig. 1B–D), and macrophages expressed IR, and not IGF1R (Supplemental Fig. 2E). We found that insulin-induced Akt2 phosphorylation (Fig. 2C) and mTORC1 activation (Fig. 2D) were significantly reduced in TEPMs from LysM^CreIgf1R^fl/fl mice, suggesting that IGF1R mediates insulin signals on the Akt2/mTORC1 axis. In contrast, Akt1 expression and phosphorylation levels were elevated in TEPMs from Akt2^−/−, LysM^CreIgf1R^fl/fl, or HFD-fed mice, supporting a distinct role of Akt1 isoform in insulin-resistant macrophages (Fig. 2G).

Akt2 kinase deficiency promotes an M2-like phenotype in macrophages accompanied by reduced M1 responses (9, 40). Conversely, human macrophages respond to LPS and other proinflammatory stimuli by Akt2 activation and production of cytokines that typify M1-type polarization (41). These observations, coupled with findings described in Fig. 1, prompted us to investigate the activation phenotype of insulin-resistant macrophages. We have found that Akt2 phosphorylation and secretion of IL-6 and TNF-α were reduced in LPS-treated insulin-resistant macrophages compared with insulin-sensitive control cells (Fig. 3A, 3B). The secretion of TNF-α from LPS-stimulated alveolar macrophages was also less in macrophages isolated from short-term HFD–fed, Akt2^−/−, and LysM^CreIgf1R^fl/fl mice compared with control mice (Fig. 3C). Expression levels of IL-12 and iNOS were also reduced in insulin-resistant macrophages treated with LPS (Fig. 3D, 3E). The ability of mitochondrial to produce ROS was also significantly lower in insulin-resistant, compared with control, macrophages 6 h after LPS stimulation, suggesting reduced bactericidal capacity (Fig. 3F). To further characterize the activation phenotype of insulin-resistant macrophages, we measured expression of arginase 1, Fizz1, and Ym1, central determinants of M2-like activation phenotype. The results showed increased levels of arginase 1, Ym1, and Fizz1 in all three insulin resistance models (Fig. 4A–D). Expression of arginase 1 was also increased in non-TEPMs; alveolar and interstitial lung macrophages isolated from short-term HFD–fed, Akt2^−/−, and LysM^CreIgf1R^fl/fl mice compared with control mice (Fig. 4E–G). Collectively, the aforementioned findings suggest that insulin-resistant macrophages acquire an M2-like phenotype (M-InsR).

The PI3K/Akt/mTORC1 pathway plays a central role in metabolic homeostasis by regulating glucose uptake and metabolic enzymes. TEPMs derived from HFD-fed mice have higher glycolytic rate compared with insulin-sensitive cells (Fig. 5A). Similarly, insulin-induced glycolysis was also abrogated in insulin-resistant macrophages (Fig. 5B). Production of lactate, the end product of glycolysis, was also elevated in the cell culture medium from all types of insulin-resistant macrophages (Fig. 5C). Expression of key metabolic enzymes participating in glycolysis was measured. The first step of glycolysis, the conversion of glucose to glucose-6-phosphate, is catalyzed by hexokinase. The mRNA of hexokinase 3 (Hk3) was significantly increased in insulin-resistant, compared with insulin-sensitive, macrophages (Fig. 5D). Similarly, expression of phosphofructokinase and lactate dehydrogenase (LDHa), catalyzing downstream steps of the glycolytic cycle, was also elevated in insulin-resistant macrophages (Fig. 5E, 5F).

Macrophages express primarily Glut1 and Glut3 to mediate glucose uptake (42). Expression of both Glut1 and Glut3 was elevated in insulin-resistant macrophages (Fig. 5G), and basal glucose uptake capacity was concomitantly increased (Fig. 5H), supporting the elevated glycolysis observed in the absence of stimulus.

To determine whether increased glycolysis is causally linked to M2-like polarization of insulin-resistant macrophages, we treated TEPMs from insulin-resistant macrophages with 2DG to inhibit glycolysis. Interestingly, inhibition of glycolysis suppressed expression of the M2-polarization markers arginase 1 and Fizz1. Similarly, inhibition of mTORC1 by rapamycin also suppressed M2-like polarization of insulin-resistant macrophages (Fig. 6A, 6B).

mTORC1 controls glycolysis through regulation of glycolytic enzymes. Akt1 deficiency and subsequent impaired mTORC1 activation resulted in reduced glucose uptake, glycolysis rate, and LDHa expression (Supplemental Fig. 3A–C). To determine whether inhibition of mTORC1 or of glycolysis reverses upregulation of glucose transport and glycolytic enzyme expression, we measured the levels of Glut1, Hk3, and LDHa following rapamycin treatment. The results showed that treatment of insulin-resistant macrophages with rapamycin or 2DG alleviated the observed increase of Glut1, Hk3, and LDHa (Fig. 6C–E) without suppressing transcription in a nonspecific manner (Supplemental Fig. 3D).

Sepsis, often complicated by lung injury, is a common etiology of morbidity and mortality. Obesity has been unexpectedly associated with improved short-term survival from sepsis, whereas increased susceptibility to bacterial infections is observed in patients with T2D (43–45). Macrophages are key mediators of sepsis and lung injury. Prompted by the reduced response of insulin-resistant macrophages to LPS (Fig. 3), we investigated the contribution of insulin-resistant macrophages to polymicrobial sepsis-induced systemic inflammation and lung injury. We used an in vivo model of polymicrobial sepsis and lung injury induced by CLP in mice bearing insulin-resistant macrophages (diet-induced obese mice, Akt2^−/− mice, and LysM^CreIgf1R^fl/fl mice). Systemic expression of the proinflammatory cytokines MIP-2, IL-6, and TNF-α were significantly lower in Akt2^−/− and HFD-fed mice, which exhibit systemic insulin resistance, whereas only serum IL-6 was reduced in LysM^CreIgf1R^fl/fl mice, which exhibit insulin resistance in myeloid cells only (Fig. 7A). In contrast, cytokine levels in the BALF were reduced, and arginase 1 levels were increased compared with control animals in all three models of mice bearing insulin-resistant macrophages (Fig. 7B, 7C). The severity of lung injury, as indicated by MPO activity (Fig. 7D) and histological evaluation (Fig. 7E), was alleviated in mice harboring insulin-resistant macrophages, with the exception of histological evaluation of tissue damage in LysM^CreIgf1R^fl/fl mice, which was not significantly improved. The discrepancy observed in LysM^CreIgf1R^fl/fl mice could be potentially due to the contribution of additional cell types, such as epithelial and endothelial cells, which also contribute to inflammatory responses in the lung by producing inflammatory mediators in response to pathogens and controlling fluid clearance, permeability, and integrity of alveoli (46–49). Thus, insulin signaling in these cell types may also be critical to disease pathogenesis. Similar to the findings observed in HFD-fed mice were evident in mice fed HFD for short term and exposed to CLP (Supplemental Fig. 4A–C), suggesting that reduced inflammation and tissue damage was not due to obesity, but rather to insulin resistance.

Although CLP-induced acute lung injury was reduced in insulin-resistant mice, survival was not improved (Fig. 7F). We thus evaluated the bacterial clearance capacity of mice bearing insulin-resistant macrophages and found higher bacterial load in the peritoneal lavage after CLP in mice harboring insulin-resistant macrophages compared with control animals (Fig. 7G). Phagocytic capacity was not affected by insulin resistance in macrophages, but was only moderately affected in the absence of Akt2, suggesting that insulin resistance may affect clearance of bacteria after phagocytosis (Supplemental Fig. 4D). Indeed, phagolysosomal fusion is controlled by mTOR (50) and may be affected in the context of insulin resistance.

Obesity and T2D are associated with altered innate immune responses. In the current study, we demonstrated that macrophages develop resistance to insulin when they are chronically exposed to high insulin levels in vitro and in vivo. A similar macrophage phenotype was observed when Akt2 kinase or IGF1R was deleted. Insulin-resistant macrophages exhibited increased glycolytic metabolism and possessed an M2-like phenotype (termed M-InsR), which was associated with reduced responsiveness to LPS and reduced inflammatory response to polymicrobial sepsis in vivo. We have shown that the M-InsR phenotype was characterized by increased mTORC1 activity, which affects glycolytic metabolism and expression of M2 polarization markers.

Insulin signals via Akt kinases and among the three Akt kinase isoforms; Akt2 is the one that primarily mediates insulin signaling (33, 34, 51). We had also demonstrated that Akt regulates C/EBPβ (52), the major transcriptional regulator of the M2 determinants Arg1 and Fizz1 (53–55). In addition, mTOR regulates C/EBPβ at the posttranscriptional level, providing a potential link between insulin/Akt/mTORC1 signals and Arg1 and Fizz1 expression.

Akt2-deficient mice are hyperinsulinemic and spontaneously develop insulin resistance (35). In this study, we demonstrated that this metabolic defect also applies to macrophages from Akt2-deficient mice. Akt2 is activated downstream of IGF1R in response to insulin, and macrophages from LysM^CreIgf1R^fl/fl mice develop insulin resistance. Thus, Akt2 and Igf1R-null macrophages provide important cell-autonomous models for the study of insulin resistance independent of diet.

In all three insulin resistance models examined (HFD, Akt2^−/−, and Igf1R^−/−), we showed that macrophages failed to activate Akt in response to insulin. Nevertheless, they had increased basal glycolysis, which was partly explained by elevated basal expression of glucose transporters Glut1 and Glut3 and increased expression of the glycolytic enzymes Hk3, phosphofructokinase, and LDHa. Increased glycolysis has been observed in M2 macrophages induced by IL-4 (26). IL-4 signals are mediated by PI3K/Akt and mTORC2 pathway (26), but the Akt isoform involved was not described. Similarly, helminth-induced, M2-polarized macrophages, also dependent on IL-4, required mTORC2 activation (56). Our results showed that insulin-resistant macrophages exhibited elevated basal mTORC1 activity and reduced response of Akt2 to phosphorylation upon insulin restimulation. Activation of Akt/mTORC1 has been reported to facilitate increased energy demands in IL-4–induced M2 macrophages (57). Inhibition of Akt using a pan-Akt chemical inhibitor regulated a subset of M2 markers (57), which we also found to be expressed in insulin-resistant macrophages. As the inhibitor used in that study blocked all Akt isoforms, no information about their distinct roles in insulin resistance could be obtained. In the present report, we demonstrated that selective inhibition of Akt2, which does not affect activation of Akt1, resulted in upregulation of M2 polarization markers, suggesting that Akt1 positively regulates M2 polarization genes, whereas Akt2 signals oppose this effect. In support of this hypothesis are earlier findings showing that deletion of Akt1 promotes M1 polarization (58). Our results also showed that increased mTORC1 activity in insulin-resistant macrophages was responsible for the expression of M2-polarizing genes because inhibition of mTORC1 by rapamycin suppressed expression of the M2 markers arginase 1 and Fizz1. Activation of mTORC1 also controlled expression of glycolytic enzymes, supporting its central role in Akt-mediated, insulin-resistant phenotype of macrophages.

Insulin resistance and T2D have been associated with several pathological conditions, most of which are linked to low-grade systemic inflammation being the result of multiple factors such as deregulation of adipokines and adipose tissue inflammation (59). Moreover, conditions such as cancer, in which M2-like macrophages (TAMs) support tumor growth (60), or atherosclerosis, in which foam cells exhibiting M2-like properties are found in atheromatic plaques (61, 62), are also associated with insulin resistance. The phenotype of peripheral macrophages in acute inflammatory responses in the context of insulin resistance remains poorly studied. Obesity is associated with increased susceptibility to bacterial infections accompanied with reduced inflammatory responses (43) as well as increased tumorigenesis and induction of TAMs (63), conditions suggesting presence of macrophages of the M2 spectrum (64). Paradoxically, in the context of acute inflammatory responses, obesity has been associated with improved short-term survival in sepsis in patients in the intensive care unit (44, 45). Although the mechanism of this so-called “obesity paradox” remains unknown, it is well documented that obese individuals have reduced responses to severe acute inflammation such as polymicrobial sepsis (65). Insulin resistance, which is characteristic in obesity-related T2D, has been associated with increased and sustained low-grade systemic inflammation (66). In contrast, septic patients develop insulin resistance, likely as an adaptation mechanism to systemic shock, which, if prolonged, also has detrimental effects. Clinically, insulin treatment is recommended in septic patients, aiming to reduce infections (67) and control blood glucose levels (68). The complexity of insulin resistance and sepsis can be partly explained by our findings showing that insulin resistance in macrophages dampens inflammatory responses but, at the same time, suppresses antibacterial responses necessary to clear pathogens in the context of polymicrobial sepsis. The discrepancy observed in the phenotype of LysM^CreIgf1R^fl/fl mice also highlighted the potential contribution of additional cell types such as epithelial and/or endothelial cells in sepsis-induced lung injury in the context of insulin resistance. Thus, patients with insulin resistance are less likely to have a severe inflammatory response in sepsis, provided that infections are controlled by antibiotics.

Overall, this study describes a novel macrophage phenotype that is observed in peripheral macrophages in the context of insulin resistance, which is under the control of IGF1R/Akt2/mTORC1 signals. Insulin-resistant macrophages obtained a unique M2-like phenotype and exhibited increased glycolysis. This phenotype can be considered an innate immune training state and may explain changes in macrophage responses and development of related pathological conditions that occur in obesity and T2D.

The authors have no financial conflicts of interest.