Leptin Signaling Suppression in Macrophages Improves Immunometabolic Outcomes in Obesity

Leptin is an adipokine produced by the adipose tissue (AT) and regulates food intake (1,2). Leptin is known to have proinflammatory effects (3). Leptin activates the STAT3 pathway as well as the phosphatidylinositol 3-kinase (PI3K)/mammalian target of rapamycin (mTOR) pathway (3–5). mTOR is an intracellular kinase that integrates nutrients signaling from growth factors present in the microenvironment (leptin, insulin, IL-2) to define the global metabolic rate and determine cellular growth, proliferation, and response (4,6). Animals deficient in leptin or leptin receptor present several metabolic and immunologic alterations.

Circulating levels of leptin are increased in obesity due to AT expansion and central leptin resistance (7,8). This is also associated with insulin resistance, which is the major risk factor for type 2 diabetes (9–11). AT inflammation is closely related to insulin resistance, and AT macrophages (ATMs) are key mediators of AT inflammation (8,12). ATMs contribute to systemic low-grade inflammation observed in obesity by releasing proinflammatory cytokines, such as TNF-α, IL-6, IL-12, and IL-1β, which impair insulin signaling in adipocytes (9,10,12, 13). ATMs are extremely plastic cells and can present a large spectrum of activation phenotypes in a context-dependent way. Two extreme macrophage activation phenotypes are the proinflammatory (M1) and tissue repair/anti-inflammatory (M2) phenotypes (8,14). In both activation states, macrophages exhibit increased glucose utilization to meet metabolic needs to fulfill their function (15,16). On inflammatory signaling, macrophages undergo intense metabolic reprogramming, favoring a metabolic shift toward glycolysis and reduced oxidative respiration (17,18). Tissue-derived factors can modulate macrophage immunometabolic profile and function. For example, chronic exposure to insulin promotes a unique macrophage phenotype—with M2-like features and increased glycolysis (19,20). Also, macrophages treated with leptin accumulate lipid droplets, which is a hallmark of macrophage commitment to an inflammatory state (21,22). Thus, tissue-derived factors, such as leptin, may modulate the immunometabolic profile of macrophages, contributing to obesity-induced systemic inflammation and insulin resistance.

Understanding how hyperleptinemia modulates macrophage metabolism and function is of paramount importance to understanding the mechanisms associated with obesity-induced disorders, such as type 2 diabetes. Here, we explored the importance of leptin effects in macrophages and the consequences for the metabolic and inflammatory outcomes of obesity. We show that leptin acts as a checkpoint for macrophage activation. It exacerbates the activation of macrophages triggered by LPS and IL-4 with metabolic and functional consequences. Moreover, peritoneal macrophages (pMacs) from obese and hyperleptinemic mice have elevated expression of proinflammatory cytokines and leptin receptor (LepR or ObR), as well as increased mTOR activity, which are not triggered in ATMs. Loss of leptin signaling in myeloid cells improves systemic glucose homeostasis in hyperleptinemic obese mice and reduces systemic low-grade inflammation. These findings support the model where tissue-derived cues modulate immune cells function and metabolism, affecting systemic inflammation in obesity, which brings focus to novel potential therapeutic targets downstream of leptin signaling.

Male C57BL6, LysMcre, ob/ob, and db/db mice at 6–12 weeks of age were obtained from the Multidisciplinary Center for Biological Investigation on Laboratory Animal Science (CEMIB). ObRLoxp (B6.129P2-Lepr^tm1Rck/J), RaptorLoxp (B6.Cg-Rptor^tm2.1Lex/J, Rap^loxP), and RictorLoxp (Rictor^tm1.1Klg/SjmJ) mice were purchased from The Jackson Laboratory. The ObR-reporter tdTomato animals were obtained from the Institute of Biomedical Sciences, University of São Paulo. The use of animals was approved by the Research Ethics Committee (Comissão de Ética na Utilização Animal: 4493-1/2017). High-fat diet (HFD) TD93075 was acquired from Harlan Teklad. Blood glucose was determined with a One Touch Basic glucometer (LifeScan).

These tests were performed after either 5 h (insulin tolerance test [ITT]) or 12–14 h (glucose tolerance test [GTT]) of fasting as previously described (23).

Perigonadal AT (pgAT) was removed, cut into small pieces in RPMI medium + 2% FBS, and digested for ∼40 min with collagenase type II (Sigma-Aldrich). Cell suspensions were then filtered through a sieve and centrifugated. The peritoneal lavage was collected with use of 5 mL PBS + 2% FBS, and the cells were separated by centrifugation for further staining for flow cytometry. Bone marrow–derived macrophages were obtained as previously described (24).

Macrophages were resuspended in PBS supplemented with 2% FBS. Surface markers were stained with monoclonal antibody for multicolor flow cytometry (Table 1) diluted in PBS + 2% FBS (1:200) for 30 min. Mitochondrial staining was performed as previously described (25). For intracellular staining, cells were stimulated with phorbol 12-myristate 13-acetate (50 ng/mL) and ionomycin (1 nmol/L) for 4 h. Brefeldin was added to the culture during the last 3 h. Cells were then permeabilized and fixed with CytoFix/CytoPerm (BD Biosciences). Cells were washed with BD Perm/Wash Buffer (BD Biosciences) to maintain the cells permeabilized. Cytokines and phosphorylated proteins were stained with monoclonal antibody (Table 1) diluted in BD Perm/Wash Buffer (1:200) for 40 min. Stained cells were washed with PBS and acquired with a BD LSRFortessa or BD FACSymphony flow cytometer (BD Biosciences) and analyzed with FlowJo software (Tree Star).

Cytokine release in supernatant after leptin (100 ng/mL) and/or LPS (100 ng/mL) was measured with ELISA using capture and detection antibodies against TNF-α, IL-6, and IL-12 (eBioscience) according to the manufacturer’s protocols.

Macrophage supernatant was analyzed on Cedex Bio Analyzer (Roche) with use of Glutamine V2 Bio, Lactate Bio, and Glucose Bio test kits according to the manufacturer’s instructions.

Total RNA was isolated and purified with RNeasy Mini Kit (QIAGEN). cDNA was synthesized with SuperScript III First-Strand Synthesis System (Invitrogen). Real-time PCR was applied with TaqMan probes (Applied Biosystems). Expression levels of each gene (Table 2) were normalized to housekeeping gene HPRT.

For real-time analysis of ECAR (extracellular acidification rate) and OCR (oxygen consumption rate), macrophages were analyzed with an XFe24 or XFe96 Extracellular Flux Analyzer (Seahorse Bioscience) according to the manufacturer’s instructions. Three or more consecutive measurements were obtained under baseline conditions and after the sequential addition of specific inhibitors/activators from the metabolic tests. For the mitochondrial stress test, we added 1 μmol/L oligomycin, 1.5 μmol/L FCCP (fluoro-carbonyl cyanide phenylhydrazone), and 100 nmol/L rotenone + 1 μmol/L antimycin A (all Sigma-Aldrich reagents). For the glycolytic stress test, we added glucose (25 mmol/L), oligomycin (2 μmol/L), and 2-deoxyglucose (2-DG) (50 mmol/L).

For carbon-tracing experiments, macrophages were stimulated for 1 h in the presence of ¹³C glucose and then treated with leptin (100 ng/mL) and LPS (100 ng/mL) for 5 h. Metabolites from cultured cells were extracted through washing of the cell layer with ice-cold (4°C) 0.9% NaCl solution and addition of 500 μL extraction solution (80:20 methanol:Milli-Q water [stored at −80°C] with 1 μg/mL norvaline and 1 μg/mL adipic acid) to cells. Macrophages were then scraped and incubated 2–5 min on ice, and the extraction mix was transferred to Eppendorf tubes. We centrifuged the samples for 15 min, 10,000 relative centrifugal force, and transferred the supernatant to a new Eppendorf tube. Dried metabolite extracts were resuspended in pyridine and derivatized with 20 mg/mL methoxyamine (sc-263468; Santa Cruz Biotechnology) for 60 min at 37°C and subsequently with N-(tert-Butyldimethylsilyl)-N-methyltrifluoroacetamide, with 1% tert-Butyldimethylchlorosilane (375934; Sigma-Aldrich) for 60 min at 80°C. Isotopomer distributions were measured with a DB-5MS GC Column in a 7890 GC system (Agilent Technologies) combined with a 5977 MS system (Agilent Technologies). Data processing, including correction for natural isotope abundance, was performed as previously described (26,27).

Electron microscopy was carried out at the Electron Microscopy Laboratory of the Institute of Biology at the University of Campinas. A cell monolayer grown on a glass coverslip was fixed with 2.5% glutaraldehyde in 0.1 mol/L sodium cacodylate and 3 mmol/L CaCl₂ buffer and incubated for 1 h on ice. After fixation, the samples were washed in 0.1 mol/L sodium cacodylate and 3 mmol/L CaCl₂ buffer and fixed with 2% osmium tetroxide in 0.1 mol/L sodium cacodylate and 3 mmol/L CaCl₂ buffer for 30 min and kept in acetate. Next, 2% uranyl was added and kept in the freezer overnight. The cells were dehydrated in ethanol on ice. The dehydrated cells were infiltrated in epon resin and placed in an oven at 60°C for polymerization for 72 h. Ultrathin cell slides were cut with a Leica Ultracut microtome, stained with 2% uranyl acetate and Reynolds citrate solution, and examined on a Leo 906 Transmission Electron Microscope (ZEISS) at an acceleration voltage of 60 kV.

We incubated 50 μg protein that was extracted in radioimmunoprecipitation assay (RIPA) buffer (Sigma-Aldrich) and resuspended in NuPAGE LDS Sample Buffer 4X (Thermo Fisher Scientific). The samples were then heated at 95°C for 5 min. The protein concentrate was submitted to electrophoresis gel in polyacrilamide gel 8% under 120 V tension and 30 mA. After the electrophoresis migration, we performed immunoblotting by transferring the proteins into a nitrocellulose membrane Hybond ECL, Amersham Pharmacia Biotech (Uppsala, Sweden). The membrane was blocked with 5% albumin solution in Tris-buffered saline with Tween (Tris-HCl 200 mmol/L, NaCl 500 mmol/L, Tween20, pH 7.5) for 1 h before the overnight incubation at 4°C with specific primary antibody (Table 3). The secondary antibody was incubated for 1 h at room temperature, and we used SuperSignal West Pico PLUS Chemiluminescent Substrate (Thermo Fisher Scientific) to obtain the images according to the manufacturer’s instructions.

pLKO.1 lentiviral vector (Addgene plasmid no. 8453) was used to clone shRNAs targeting the genes of interest within the AgeI/EcoRI sites at the 3-end of the human U6 promoter. The targeted sequences were as follows: GFP 5′CAAGCTGACCCTGAAGTTCAT-3′, STAT3 5′-CGACTTTGATTTCAACTACAA-3′. Bone marrow–derived macrophages (BMDM) were transduced with lentiviral particles (multiplicity of infection 1) from the pLKO shGFP or pLKO shSTAT3 on the fourth day of differentiation in the presence of L929. On the fifth day, the cells received the selection antibiotic and were maintained with 1 μg/mL puromycin for another 5 days. Then, the cells were treated with 100 ng/mL leptin and/or 100 ng/mL LPS (6 h) or 20 ng/mL IL-4 (24 h) for gene expression assays. The knockdown was quantified with quantitative PCR (qPCR) with target primers for STAT3.

A total of 4,600 cells/mm² of BMDM were seeded and differentiated with L929. On the seventh day, cells were treated with 100 ng/mL leptin for 30 min. Next, wounds were created with 200 μL pipette tips and the wells were washed once with PBS. The cells were treated with 20 ng/mL IL4 in RPMI medium supplemented with 10% FBS and immediately imaged for 16 h with use of Operetta (PerkinElmer Life Sciences) in bright-field mode every hour at 37°C in a 5% CO₂ atmosphere. The images were processed with Fiji-ImageJ using a macro based on previous work (28).

The data were analyzed with Student t test or one-way ANOVA, followed by the Tukey test or two-way ANOVA followed by the Holm-Sidak method to determine the significance of the individual differences. Data were analyzed with the statistical software SigmaStat 3.1 (Systat Software, Point Richmond, CA). Only male mice were used; thus, sex was not considered as a factor in the statistical analysis of the data.

Further information, reagents, and all other supporting data in this study are available on request from corresponding author Pedro M. Moraes-Vieira.

We found that leptin treatment prior to LPS activation led to increased mRNA expression and secretion of the proinflammatory cytokines TNF-α, IL-12, and IL-6 (Fig. 1A and B). Leptin treatment of LPS-activated macrophages from wild-type (WT) and leptin-deficient (ob/ob) mice led to enhanced LPS-induced cytokine production in macrophages (Supplementary Fig. 1B and C). Leptin had no additive effect of LPS-induced changes in pMacs and BMDM from leptin receptor–deficient (db/db) mice (Supplementary Fig. 1A–C). We observed that, in ob/ob macrophages, IL-6 levels are higher after LPS treatment compared with WT but followed a trend similar to that of BMDMs from WT mice (Supplementary Fig. 1A). Thus, leptin signaling through ObR is sufficient to enhance the production of proinflammatory cytokines in activated macrophages.

We next evaluated real-time metabolic changes in leptin-treated macrophages. We found that leptin exacerbates LPS-induced glycolysis, observed with ECAR, which was maintained even after macrophages were submitted to a glycolytic stress test (Fig. 1C and D). Leptin had no additive effect on LPS-induced blockade of oxygen consumption by macrophages (Fig. 1E and F). We next investigated whether the inflammatory effects of leptin in macrophages are dependent on glucose metabolism. We used an approach of energetic carbon source limitation. We cultivated macrophages in complete media, glucose-free media + galactose, or glucose-free media (PBS) for 6 h. This enables us to determine the importance of glycolysis to the inflammatory response. For galactose to enter the glycolytic pathway, there is energetic involvement in the form of two UDP-dependent reactions (galactose 1 phosphate formation and epimerization of UDP-galactose) (29–31). We found that the additive effect of leptin in LPS-activated macrophages was lost in glucose-deprived conditions (Gal and PBS), as measured by TNF-α and IL-6 (Fig. 1G). Thus, leptin activity in macrophages requires biomolecules generated by glucose metabolism to fuel its downstream pathways and promote inflammation. Next, we determined the significance of metabolic alterations to the function of leptin-treated macrophages. We did a pharmacological inhibition of lactate dehydrogenase (LDH) using oxamate (OXA) and observed that BMDM failed to induce leptin-dependent cytokine expression (Il6) on oxamate treatment, with no changes in Tnfα expression levels (Supplementary Fig. 1D). Our results show that LDH activity is essential for leptin-dependent booster in LPS-induced proinflammatory cytokine expression. We then pharmacologically inhibited mitochondrial pyruvate carrier with UK-5099, preventing pyruvate entry into mitochondria without affecting glycolysis. UK-5099 treatment did not have the same inhibitory effect in proinflammatory cytokine expression triggered by leptin in LPS-activated macrophages as observed with OXA (Supplementary Fig. 1E). Thus, pyruvate entry, and its subsequent oxidation in mitochondria, is not necessary for leptin-induced booster of proinflammatory cytokine expression on LPS activation. Together, these results indicate that glycolysis, but not mitochondrial metabolism, is important for leptin effects in activated proinflammatory macrophages.

During the metabolic adaptation, LPS-activated macrophages shut down fatty acid oxidation and present breaks in the tricarboxylic acid cycle that allow the accumulation of succinate and citrate and production of itaconate (15,32). Using ¹³C glucose, we did not observe further carbon incorporation into citrate and itaconate by leptin in relation to LPS-activated macrophages (Supplementary Fig. 1F and G). This indicates that the metabolic rewiring promoted by LPS is sufficient to induce increased cytokine production through stimulation with leptin even without additive effects on the inhibition of oxygen consumption and alterations in the tricarboxylic acid cycle. However, because changes in glycolytic rates are related to mitochondrial modulation, we next analyzed the contents of subunits representing the electron transport chain (ETC) mitochondrial complexes and observed that LPS-induced reduction in ETC complexes is exacerbated by cotreatment with leptin (Fig. 1H). We then investigated the mitochondrial membrane potential in leptin-treated macrophages, since mitochondrial polarization can reflect the functionality of the ETC. Analysis of mitochondrial phenotype using the probes MitoTracker Green and Red revealed an increase in depolarized mitochondria (MitoGreen^high MitoRed^low), whereas hyperpolarized mitochondria (MitoGreen^high MitoRed^high) were decreased in leptin-treated macrophages (Fig. 1I), even though the total mitochondrial mass remained equivalent (Supplementary Fig. 1H). Moreover, ultrastructural analysis showed that, consistent with elevated ECAR and reduced levels of mitochondrial ETC complexes, LPS and leptin cotreated macrophages display enhanced alterations in mitochondria morphology, observed as short and rounded mitochondria with looser cristae (Fig. 1J). These results indicate that leptin contributes to the inflammatory phenotype through potentiation of LPS effects on mitochondrial morphology and function, and glucose-dependent metabolism.

Leptin signaling in macrophages increases LPS-induced cytokine release through metabolic changes. We used the Cre-lox system to generate mice with a specific deletion of ObR in myeloid cells. We used ObR floxed (ObR^fl/fl) and LysM^Cre mice to study macrophages lacking leptin signaling. Consistent with what we observed from db/db mice, mice with conditional deletion of leptin receptor (ObR^−/−) in macrophages fail to enhance LPS-induced expression and secretion of proinflammatory cytokines (TNF-α, IL-12, IL-6) compared with control (ObR^fl/fl) (Fig. 2A and B). We also analyzed the percentage of TNF-α–positive and IL-12–positive macrophages and IL-6 secretion in LPS-activated macrophages from both ObR^fl/fl and ObR^−/− mice and found that lack of leptin receptor impaired the enhanced production and secretion of TNF-α, IL-12, and IL-6 in response to leptin plus LPS (Fig. 2B and C and Supplementary Fig. 2A).

To determine the relevance of signaling pathways responsible for leptin-mediated effects on macrophages, we analyzed the downstream activation of mTOR by leptin. We observed that leptin induced the phosphorylation of both mTORC1 and mTORC2 downstream targets, 4EBP1 and AKT^Ser473, respectively (Fig. 2D). Cotreatment with leptin and LPS sustained the phosphorylation of AKT^Ser473 rather than 4EBP1 (Fig. 2D). We also found that S6 kinase, an mTORC1 target, is phosphorylated on LPS activation, whereas cotreatment with leptin did not alter the signal intensity, showing that LPS-induced mTORC1 activation was not magnified by leptin (Supplementary Fig. 2B). Rapamycin is a potent mTOR inhibitor that when administered at higher doses inhibits both complexes of mTOR (mTORC1 and mTORC2) (33–36). We treated macrophages with 200 nmol/L of rapamycin for 20 min before activating the cells with leptin/LPS to inhibit both mTORC1 and mTORC2 pathways. Rapamycin-treated cells displayed reduced phosphorylation of 4EBP1 and AKT^Ser473, indicating inhibition of both mTORC1 and mTORC2 (Fig. 2D). Thus, mTOR complexes are differentially activated by leptin after LPS treatment.

Next, we determined the effect of rapamycin treatment on leptin-mediated cytokine secretion in activated macrophages. Rapamycin also limited the exacerbated cytokine production by macrophages cotreated with leptin and LPS, reducing the levels of cytokines back to those induced by LPS alone (TNF-α and IL-12) and even to control levels (IL-6) (Fig. 2E). To determine whether mTOR participates in leptin-induced metabolic adaptation of macrophages, we also treated BMDMs with rapamycin prior to activation with leptin and LPS and submitted the cells to a glycolytic stress test. We found that mTOR signaling is necessary for the leptin-dependent booster in glycolytic capacity (Fig. 2F). The STAT3 pathway can also be triggered by leptin. Next, we investigated the contribution of the STAT3 pathway for leptin effects on macrophages. First, we used shRNA to target STAT3 (shSTAT3). We confirmed STAT3 knockdown in BMDM prior to the experiments (Supplementary Fig. 2C). Leptin-dependent booster of LPS-induced Tnfα expression still occurred in STAT3-deficient macrophages. (Supplementary Fig. 2C). Together, these results indicate that leptin-induced mTOR activation, and not the activation of STAT3 pathway, is necessary for leptin-mediated effects on LPS-activated macrophages.

To better understand the role of each mTOR complex in leptin-induced macrophage activation, we crossed Rictor and Raptor floxed mice (Rictor^fl/fl and Raptor^fl/fl) with LysM^Cre mice to generate macrophages lacking mTORC2 and mTORC1 activity, respectively. Rictor and Raptor are scaffold proteins that integrate mTORC2 and mTORC1 complexes, and they are necessary for the complex to be functional (37). TNF-α and IL-6 levels were significantly reduced in leptin/LPS-treated macrophages derived from Rictor knockout (Rictor^−/−) mice compared with control Rictor^fl/fl (Fig. 2G), which was similar to what we observed in ObR^−/− macrophages (Fig. 2A–C). We did not observe a significant difference in TNF-α and IL-6 concentration in the supernatant of either Raptor^−/− or Raptor^fl/fl control macrophages after LPS and leptin plus LPS treatments (Fig. 2H). Together, our data indicate that mTOR activity fuels the exacerbated inflammatory response mediated by leptin, specifically, through the mTORC2 pathway.

Next, we characterized the role of leptin signaling in macrophages in an in vitro condition of disrupted metabolic homeostasis to mimic what happens in vivo during obesity. We performed in vitro metabolic activation (Mme) of macrophages (addition of glucose 30 mmol/L, insulin 10 nmol/L, and palmitate 0.4 mmol/L) as previously described (38). We found that the lack of leptin signaling in Mme macrophages increased the expression of the M2-related markers Il10 and Mrc1 (Supplementary Fig. 2D). However, as the mice gain weight, other systemic factors enable an accentuated inflammatory environment. In this context, leptin favors the proinflammatory phenotype. We also observed that Mme macrophages from ObR^−/− mice displayed increased expression of lipid metabolism markers Cd36 and Abca1 with no effect on Plin2 (Supplementary Fig. 2E). This suggests that the macrophages become dysregulated and the loss of leptin signaling indulges the dysfunctional response.

Considering the importance of mTORC2 for the differentiation of M2-like macrophages (IL-4) (39) and that mTORC2 is necessary for the effects of leptin in activated macrophages, we hypothesized that leptin may act as a systemic nutritional status signaling molecule to modulate macrophage phenotype regardless of their inflammatory milieu, since both classical and alternative macrophage activation processes are energy demanding. Thus, we polarized macrophages into a proinflammatory (IFN-γ + LPS)-like or repair (IL-4)-like profile. Using leptin receptor reporter mice, we observed that IL-4–treated macrophages [M(IL-4)] showed increased ObR expression in comparison with resting (M0) and proinflammatory [M(IFN-γ + LPS)] macrophages (Fig. 3A). We also performed a time course analysis of ObR expression in macrophages treated with either IL-4 or LPS for 3, 6, and 24 h. We found that M(IL-4) express the highest levels of ObR 6 h after the stimulus (Fig. 3B). Next, we isolated ATMs from ObR reporter mice and determined ObR expression in CD11c⁺ and CD206⁺ macrophages. CD11c is expressed predominantly in proinflammatory ATMs and CD206 in tissue-resident/anti-inflammatory ATMs (8,12,40). Similar profile was observed in ATMs, where CD206⁺ macrophages showed the highest ObR levels (Fig. 3C).

We then hypothesized that M(IL-4) would also be susceptible to leptin-induced metabolic adaptations, due to the elevated ObR expression. We analyzed the total glucose, lactate, and glutamine contents in the supernatant of M(IL-4) and observed a reduction in glucose concentration, indicating higher glucose consumption by these cells (Supplementary Fig. 3A), with no change in lactate and glutamine levels (Supplementary Fig. 3A). We observed that cotreatment with leptin and IL-4 enhanced IL-4–mediated reduction of glycolysis and glycolytic capacity and increased OCR and maximal respiratory capacity (Fig. 3D and E and Supplementary Fig. 3B and C). We next analyzed the expression of a series of alternatively activated macrophage markers (Arg1, Mrc1, Il10, Ym1, Mgl1/2). We observed that leptin boosted the expression of all M(IL-4)-associated markers (Fig. 3F). We then performed a real-time migration assay to assess the role of leptin in an in vitro model of macrophage-induced tissue repair, which enables us to determine the function of IL-4–polarized macrophages. We observed that loss of leptin signaling in M(IL-4) reduces macrophage migration into the wound site (Fig. 3G). We also showed that this is context dependent, since in the absence of IL-4 stimulus, leptin does not affect cell migration, as observed in the same migration rates in the WT and ObR^−/− BMDM (Fig. 3H). Thus, following along with our hypothesis, leptin cannot enhance the repair phenotype by itself, but when macrophages are primed with IL-4 and potentially in an IL-4–rich microenvironment, leptin signaling boosts the tissue repair phenotype. Together, our data show that leptin functions as a checkpoint to boost macrophage activation and polarization, amplifying metabolic and immunologic alterations to support the desired phenotype for which macrophages have been primed.

Finally, to better understand the role of mTOR in M(IL-4) metabolic adaptation, we submitted BMDMs from WT, Rictor^−/−, and Raptor^−/− mice to a Mito Stress Test. Raptor-deficient BMDMs showed lower OCR, as well as ECAR, compared with WT (Supplementary Fig. 3D and E). Maximal respiration was partially restored by leptin treatment in Raptor^−/− cells, albeit still reduced in comparison with control (Supplementary Fig. 3F). On IL-4 treatment, we observed that Raptor deletion inhibited glycolysis in macrophages, which was not affected by leptin treatment (Supplementary Fig. 3D). We also observed that, much like Raptor-deficient BMDM, Rictor^−/− macrophages also have lower metabolic rates, as characterized by OCR and ECAR (Fig. 3G and H). Moreover, we analyzed the importance of STAT3 in leptin-dependent increase in the expression of M2 marker Mrc1. We found that loss of STAT3 signaling on M(IL-4) reduced the expression of Mrc1 in comparison with control shGFP. However, leptin-mediated booster of Mrc1 in shSTAT3 BMDMs remained intact (Supplementary Fig. 3I). This highlights the importance of mTOR activity to optimal cellular function, since macrophages adapt their metabolism during the differentiation process. mTOR is necessary for the metabolic reprogramming during IL-4 activation (39). We suggest that leptin acts, at least in part, in a mechanism that is independent of mTORC1 but dependent on mTORC2 to compensate for the energetic demand in M(IL-4).

Because obesity is associated with changes in macrophage phenotype, we used a model of diet-induced obesity, which also leads to hyperleptinemia and inflammation (41). This is an excellent model where macrophage phenotype and increased leptin levels can be analyzed simultaneously in vivo. Because leptin enhances both proinflammatory macrophage activation and tissue repair/anti-inflammatory profile in vitro and both leptin and proinflammatory macrophages are increased in the AT of obese people and mice, we hypothesized that inhibition of leptin signaling on macrophages could directly impact systemic low-grade inflammation and the subsequent insulin resistance. We first submitted ObR^−/− mice to HFD. Control (ObR^fl/fl) and ObR^−/− mice had no changes in body weight or weight gain (Supplementary Fig. 4A). Next, we analyzed systemic insulin resistance and glucose tolerance. ObR^−/− mice showed improved glucose tolerance compared with control ObR^fl/fl, as shown by faster reduction in blood glucose levels and reduced area under the curve (Fig. 4A and B). ObR^−/− mice also displayed improved insulin sensitivity on HFD feeding (Fig. 4C and D). When fed a normal diet, ObR^fl/fl and ObR^−/− mice showed similar glucose tolerance and insulin sensitivity (Supplementary Fig. 4B–E). Moreover, we observe that in both ObR^fl/fl and ObR^−/− mice, basal glucose levels in the serum are not different after 5 h or 12 h of fasting (Supplementary Fig. 4F). Only after hyperglycemic stress (GTT) were we able to determine the faster recovery ObR^−/− HFD-fed mice (Fig. 4A). Glucose and cholesterol levels increase in both ObR^fl/fl and ObR^−/− HFD-fed mice under nonfasting conditions and cholesterol levels are decreased in HFD ObR^−/− mice in comparison with WT HFD-fed mice (Supplementary Fig. 4G). This suggests that loss of leptin signaling in myeloid cells of obese mice is important to improve systemic glucose homeostasis, as observed when mice were submitted to a stress test (GTT and ITT). Thus, impairment of leptin signaling in myeloid cells is sufficient to improve systemic metabolic impairment caused by HFD feeding.

Because inflammation has a causative role in obesity-induced insulin resistance (12,42), we evaluated the macrophage phenotype in lean and obese ObR^fl/fl and ObR^−/− mice. There were no significant differences in the production of proinflammatory cytokines by ATMs isolated from pgAT of obese ObR^fl/fl and ObR^−/− mice (Fig. 4E). However, pMacs from ObR^−/− mice produced lower levels of IL-6 and IL-1β compared with control (ObR^fl/fl) and similar levels of TNF-α (Fig. 4F), which can directly contribute to systemic low-grade inflammation (9). ATMs and pMacs from chow-fed ObR^fl/fl and ObR^−/− mice produced similar levels of proinflammatory cytokines (Supplementary Fig. 4I–J).

We analyzed the expression of proinflammatory (M1) markers and repair (M2) markers in both ATMs and pMacs of ObR^fl/fl and ObR^−/− mice. We found that, in ATMs, the frequency of proinflammatory CD11c⁺ cells increases when both ObR^fl/fl and ObR^−/− mice are fed an HFD (Fig. 4G). Interestingly, though, the frequency of M2/repair-related marker CD206 is only decreased after HFD in ATMs of ObR^fl/fl mice (Fig. 4G). ObR^−/− mice maintained a similar frequency of CD206⁺ ATMs during normal diet (ND) and HFD feeding (Fig. 4G). This affects the ratio of proinflammatory to repair types of macrophages, where the CD11c-to-206 ratio is elevated in ObR^fl/fl mice after HFD but not in ObR^−/− mice (Fig. 4H). We observed no significant changes in the frequency of CD301⁺ ATMs between the groups (Fig. 4G and H). The CD11c-to-CD301 ratio was higher in both ObR^fl/fl and ObR^−/− mice after HFD (Fig. 4H). In pMacs, we found an increase in MHCII in ObR^−/− mice even during ND, but no other significant changes were observed regarding the frequency of CD206⁺ or CD301⁺ pMacs between the groups (Fig. 4I). There were also no significant changes in the ratios of MHCII to CD206 or MHCII to 301 pMacs on either diet (Fig. 4J). Our data highlight how differently these macrophages respond to nutritional alterations. The lack of leptin receptor prevents the raise in proinflammatory-to-repair ratio caused by HFD feeding, which agrees with our data that show no significant increases in the expression of proinflammatory cytokines (TNF-α, IL-6, IL-1β) in ATMs of ObR^−/− obese mice.

Next, we aimed to determine the mechanisms by which ATMs and pMacs from obese mice respond differentially. We submitted WT mice to HFD feeding for 8 weeks. HFD-fed mice presented increased weight gain, elevated fasting glucose, and elevated levels of leptin both in serum and peritoneal lavage (Fig. 4K–M). We then assessed the pattern of ObR expression in vivo from macrophages of various tissues. Using ObR reporter mice, we found that pMacs express the highest levels of ObR compared with other tissue-resident macrophages, such as ATMs from pgAT, brown AT, subcutaneous AT, and splenic macrophages (Fig. 4N). The expression of ObR in pMacs and ATMs was also not affected by HFD (Supplementary Fig. 4K). Thus, ObR expression in these tissue-resident macrophages seems to be modulated by the microenvironment in which they reside. In the obese AT, macrophages are exposed to leptin in high concentrations, whereas the peritoneal cavity is not rich in leptin. Accordingly, the levels of leptin in the peritoneal cavity are ∼10 times lower than in the blood (peritoneal cavity, lean mean ± SEM 1,167 ± 93.64, obese 2,749 ± 225.7; serum, lean 3,518 ± 728.9, obese 33,629 ± 5,395) (Fig. 4K and L). Thus, the reduced expression of ObR in ATMs may prevent hyperresponsiveness to leptin in AT. It is likely that pMacs are more prone to intensify leptin responses because of higher expression of ObR, since this intensified response is abolished in ObR^−/− mice (Fig. 4N). Once again, these data highlight the importance of ObR expression in pMacs and the subsequent activation of mTOR by leptin to promote functional changes even if the macrophage phenotype remains similar. Our results indicate that blocking leptin signaling (ObR^−/−) in myeloid cells ameliorates systemic glucose homeostasis and inflammation in obesity.

The expression of leptin receptor in pMacs and ATMs and the levels of leptin in the microenvironment where these cells reside suggest that they may respond differentially to leptin signaling, which could directly impact their responsiveness to an inflammatory stimulus. Our next step was to evaluate the activity of mTOR complexes in vivo in conditions where leptin levels are increased, such as during obesity. We submitted mice to HFD and isolated pgAT and pMacs from both lean and obese mice (Supplementary Fig. 5A) and analyzed cytokine production as a readout of the obesity-induced inflammatory state. We observed elevated cytokine production in ATMs (TNF-α) and pMacs (TNF-α and IL-6) of obese hyperleptinemic mice (Fig. 5,A and B), consistent with our in vitro data. We also selected CD206⁺ ATMs and pMacs and noticed that in ATMs, only IL-6 was increased during HFD feeding, whereas CD206⁺ pMacs showed elevated levels of both TNF-α and IL-6 (Supplementary Fig. 5B and C). Next, we analyzed the phosphorylation of STAT3, of S6 kinase, and of AKT^Ser473 in ATMs and pMacs. Both ATMs and pMacs from HFD-fed mice presented higher levels of phosphorylated STAT3 (p-STAT3), as a possible consequence of hyperleptinemia and intense leptin signaling in these macrophages (Fig. 5C and D). However, ATMs and pMacs showed a distinct mTOR activation pattern. The phosphorylation of S6 kinase and AKT^Ser473 (p-S6 and p-AKT, respectively) was diminished in ATMs and increased in pMacs (Fig. 5C and D). A similar pattern was found in CD206⁺ macrophages from the pgAT and peritoneal lavage of obese mice (Supplementary Fig. 5D and E). These results indicate that leptin levels in the microenvironment and ObR expression in tissue-resident macrophages affect the activation of the mTOR pathway in a tissue-specific manner during obesity.

The contribution of immune cells to obesity-related inflammation has been extensively reported (43,44). Leptin deficiency promotes an anti-inflammatory profile in mast cells, which improves obesity and insulin resistance in mice (45). Even though hyperleptinemia contributes to low-grade systemic inflammation, it is still not entirely clear how leptin modulates macrophage inflammation in vivo. Here, we showed by genetic deletion of ObR in myeloid cells that blocking leptin signaling in macrophages is sufficient to improve glucose tolerance and insulin sensitivity in obese mice. We also showed that pMacs play a key role increasing systemic inflammation because of their phenotypical features, which include high expression of ObR, elevated response to leptin, and increased expression of proinflammatory cytokines. Our work highlights the role of leptin signaling in tissue-resident macrophages in obesity and the differential contribution of mTOR complexes to the hyperinflammatory response. Moreover, we propose the concept that leptin acts as a metabolic and systemic nutritional checkpoint for macrophage response.

Several studies in the literature analyzed the acute effects of leptin on the inflammatory response. Pini et al. (2008) (46) showed that acute leptin injection exacerbated the inflammatory cell infiltrate in ob/ob mice using a zymosan-induced model of peritonitis. Conversely, other reports identified acute leptin injection as an important mediator of neutrophil migration and pMacs activation (21,47). Moreover, in another study, Bernotiene et al. (2004) (48) showed that in ob/ob and db/db mice, loss of leptin or leptin signaling delayed the resolution phase of acute inflammation. They also suggest that the chronic leptin deficiency impairs the control of a proper immune response. This is also evidenced by another work where leptin-deficient mice had increased graft survival due to reduced inflammatory allogeneic response (49). The effect of chronic hyperleptinemia in macrophage activation in obesity is only evident as the mice gain weight and develop metabolic syndrome and endotoxemia (50). Thus, with our in vitro model we were able to characterize the role of leptin-induced macrophage response in an inflammatory environment, such as those induced by obesity (low-grade systemic inflammation). On Mme, macrophages become dysregulated and the loss of leptin signaling nourished the dysfunctional response. Thus, loss of leptin signaling in myeloid cells can improve systemic glucose levels, but during in vitro metabolic activation, similar to what we observed in ATMs in obesity, macrophages become dysregulated and apparently react to the metabolic alterations in the environment in a leptin-independent way.

Circulating LPS leads to activation of tissue-resident immune cells (51). Moreover, the development of obesity and obesity-enhanced gut permeability and the subsequent circulating LPS alter the macrophage population in the AT, leading to increased cell infiltration and imbalance in the pro- and anti-inflammatory phenotypes (M1- and M2-like macrophages) (12,51). AT inflammation and immune dysregulation observed in obese mice result in insulin resistance (3,12,52,53). The majority of previous studies on leptin focused on the systemic effects of this adipokine in metabolism and immunity (ob/ob and db/db mice) (3,12,52,54–58). These mice display several systemic alterations that can modulate the immune response, and the effects of leptin in these contexts may be indirect, as both of these mouse models (ob/ob and db/db) are hyperphagic and obese and can develop diabetes, hyperlipidemia, and metabolic endotoxemia (59). Previous reports have shown that LPS induces an increase in circulating levels of leptin, as well as upregulates leptin expression in the AT (60–62). These changes in leptin production are associated with elevated cytokine levels (TNF, IL-1) and anorexia (62). Obesity is also a major cause of elevated circulating LPS (50,60). As we have shown here, hyperleptinemia profoundly alters LPS-activated macrophage metabolism and response. Moreover, LPS injection in leptin-deficient mice (ob/ob) leads to high lethality rates, which can be reversed by leptin treatment (61). This study also shows that leptin deficiency reduces the production of anti-inflammatory cytokines, highlighting the extensive roles of leptin as a pleiotropic adipokine in the immune response (61). When primed to a repair phenotype (IL-4), leptin improves macrophage wound healing capacity that is accompanied by increases in metabolic rates, whereas, when primed with LPS, leptin exacerbates the proinflammatory cytokine production in a glycolysis-dependent way. Leptin can also be produced and released by immune cells, such as regulatory T cells, and plays a negative autocrine role to inhibit the expansion of regulatory T cells (63). In this context, leptin production by the AT and immune cells orchestrates systemic immune response and metabolic changes. The duality of the leptin role in immune activation is deeply rooted in the environmental signals and different cell types that respond via ObR, as observed in the adaptability of ATMs during obesity and the promotion of inflammation by pMacs. Leptin signals in the hypothalamus through activation of the JAK/STAT3 pathway. The majority of the effects of leptin on energy expenditure and glucose homeostasis are described in the context of proopiomelanocortin neuron activation via STAT3, which are anorexigenic neurons that control feeding behaviors (64). In immune cells, there is evidence that for the functional adaptation mediated by leptin treatment, the PI3K/mTOR pathway plays a major role (21). Conversely, we show that lack of STAT3 signaling does not affect leptin-mediated booster in macrophage activation in the same way we observe by inhibiting mTOR activity. Unlike hypothalamic neurons, immune cells are more susceptible to sudden changes in nutrient availability, inflammatory markers (endotoxemia, chemokines, cytokines), and direct interaction with other cell types (12,60). Therefore, it is possible that leptin activates distinct intracellular pathways in different cell types depending on the immunometabolic signal and the appropriate types of response that it demands. Additionally, since the expression of ObR can be modulated by such environmental cues, researchers face an extra challenge, to pin down the beneficial and detrimental roles of leptin.

The obesogenic environment influences ATM inflammatory state (12,65). Hyperleptinemia can be observed in mice 4 weeks after starting chronic HFD feeding. Boutens et al. (66) showed that the AT environment is responsible for ATM metabolic rewiring, specifically, the induction of glycolysis that contributes to ATM proinflammatory trait in obesity. They showed that this process does not depend on leptin alone (66). Here, we show that this appears to occur through a mechanism that involves reduced leptin signaling due to reduced ObR expression in ATMs, as ObR-deficient ATMs from obese mice have the same proinflammatory characteristics as WT ATMs. We suggest that the mechanism underlying systemic inflammation in obesity is not restricted to increased AT inflammation but, rather, involves other tissue-resident cells where ObR expression and signaling are elevated, such as in macrophages residing in the peritoneal cavity.

Leptin potentiated the activation of macrophages by LPS and IL-4. When primed for a specific profile, leptin acts as a checkpoint enhancing immunometabolic features that determine macrophage phenotype. We found that mTOR activation by leptin is necessary for the hyperinflammatory state of activated macrophages. Also, Rictor^−/− macrophages presented the same response as ObR^−/− cells, indicating that downstream activation of mTORC2 by leptin signaling is sufficient to modulate the inflammation in macrophages. The importance of mTORC2/AKT activation by leptin in bacterial cleansing was also recently shown (67). Genetic deletion of ObR in macrophages reduces proinflammatory cytokine expression, allows for AKT^Ser473 dephosphorylation, and reduces lysosomal function (67). In a systemic inflammatory condition, such as obesity, impairments in the leptin/mTOR pathway lead to overall improved glucose homeostasis.

Leptin is an abundant hormone present in homeostatic conditions favoring resident and repair phenotypes in macrophages. Under pathological stimuli, aberrant concentrations of leptin in an inflammatory environment favor the proinflammatory phenotype and response of macrophages, which can be directly reflected in the outcome of disease severity and life expectancy. The results of this study raise the possibility that targeting leptin downstream pathways has a great potential for the development of novel therapeutic approaches to treat obesity-induced inflammation and type 2 diabetes. Personalized therapeutic approaches that take into consideration the individual needs of each patient are growing in clinical practices. Due to the different aspects of leptin in the modulation of immune response, targeted therapies for conditions that lead to systemic metabolic disruption (cachexia, anorexia, obesity) should be considered. The use of full knockout models (ob/ob and db/db) did not allow researchers to fully understand the contribution of leptin signaling in macrophages and other immune and even nonimmune cell types. For example, in ob/ob mice, reconstitution of leptin improves metabolic parameters simply because the mice regain their feeding control (behavior) (58). The use of cell-specific knockdown models makes it possible to investigate the effects of leptin on macrophages, reducing the influence of leptin action on other cell types, such as leptin-central actions on energy expenditure, and nutrient availability through fast/feeding. Here, we have shown that selective knockout of leptin receptor in myeloid cells ameliorates the deleterious effects of obesity. Because macrophage activation contributes to systemic low-grade inflammation, limiting leptin signaling in macrophages via selective delivery systems might help reduce inflammation and restore glucose homeostasis without interfering in other cell types.

Acknowledgments. The authors thank the Metabolomics Core at the Max Planck Institute of Epigenetics and Immunobiology for advice and support of this project.

Funding. This work was supported by the São Paulo research foundation Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP) (2013/07607-8, 2015/15262-8, 2016/23328-0, 2019/25973-8, 2019/19435-3, 2018/19338-5, 2019/06372-3, and 2020/16030-0). The authors are also thankful for the support of Max-Planck Society, the Brazilian National Council for Scientific and Technological Development (CNPq), and Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES), Finance Code 001.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Authors Contributions. L.d.B.M. designed, performed, and analyzed experiments, designed figures, and wrote the manuscript. J.S.P., C.F.d.A., F.C.-d.-S., A.C., N.v.T.B., G.G.D., B.C., J.A.d.S.P., J.C., L.M.d.R., G.C., G.R., J.v.V.-d.S., D.A., and S.M.G.D. designed and performed experiments. J.B. designed experiments and analyzed data. S.R.C., J.D., E.J.P., and N.O.S.C. designed experiments and provided reagents. P.M.M.-V. supervised the study, designed experiments, and wrote and edited manuscript. P.M.M.-V. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.