Adipose Tissue Macrophages Function As Antigen-Presenting Cells and Regulate Adipose Tissue CD4⁺ T Cells in Mice

C57BL/6J, Cx₃cr1^GFP/+, and OT-II mice were from The Jackson Laboratory. Male mice (6 weeks old) were fed a normal diet (ND; 4.5% fat; PMI Nutrition International) or HFD (60% fat; Research Diets). All mice procedures were approved by the University of Michigan University Committee on Use and Care of Animals and were in compliance with the Institute of Laboratory Animal Research Guide for the Care and Use of Laboratory Animals.

Gene expression analysis was performed as described (38). Relative expression in duplicate samples was determined using the 2^–ΔΔCT method after normalizing to Arbp. Sequences for PCR primers are provided in Supplementary Table 1.

Fat samples and cultured cells were prepared for immunofluorescence microscopy as described (6,39). Images were captured on an Olympus inverted microscope with an Olympus DP72 camera. Confocal images were captured with a FluoView microscope and software (Olympus) or on a Zeiss LSM 700 microscope and software.

Stromal vascular cells (SVCs) were isolated as described (38). SVCs were plated at 1 × 10⁶ cells/mL in DMEM containing 10% heat-inactivated (HI) FBS.

Cells were stained for flow cytometry as described (38). Antibodies for flow cytometry are provided in Supplementary Table 2. Live/dead fixable violet staining kits (Molecular Probes) and Foxp3 staining kits (eBiosciences) were used according to the manufacturers’ instructions. Cells were analyzed on a FACSCanto II Flow Cytometer (BD Biosciences) using FlowJo 9.4 software (Treestar). SVCs (5 × 10⁷/mL) were suspended in RPMI/2%HI-FBS containing propidium iodide (2 μg/mL) and sorted on a FACSAria (BD Biosciences). Fluorescence-activated cell sorted (FACS) cells were cultured in DMEM/10%HI-FBS or lysed in RLT buffer (Qiagen).

BrdU (0.8 mg/mL; Sigma) was added to the drinking water. Isolated cells were stained with indicated cell-surface antibodies and permeabilized using Foxp3 staining kits. Permeabilized cells were incubated in DNAseI (30 μg; Sigma) at 37°C for 1 h and stained with BrdU and Foxp3 antibodies.

Fluorescein isothiocyanate (FITC)-labeled ovalbumin (OVA; 100 μg/mouse; Molecular Probes) was injected intraperitoneally into mice 1 h before fat pads were examined by immunofluorescence microscopy. SVCs were cultured on glass chamber slides (Laboratory-Tek) overnight before FITC-labeled OVA (100 μg/mL; Molecular Probes) or FITC-labeled ZymosanA (Saccharomyces cerevisiae) bioparticles (100 particles/cell; Molecular Probes) were added to the media for 1 h. Lean mice were injected intraperitoneally with DQ-OVA (100 μg/mouse; Molecular Probes), a self-quenched conjugate of OVA and boron-dipyrromethene (BODIPY) that fluoresces upon proteolytic cleavage within endolysosomes (40). Peritoneal macrophages and SVCs were isolated 1 h later and stained for flow cytometry.

SVCs or FACS-sorted cells were grown in 96-well U bottom plates overnight, then pulsed with medium alone, OVA peptide (OVA_323-339; Abbiotec), or whole OVA (ThermoScientific) for 2 h at 37°C. CD4⁺ T cells were isolated from the spleen and lymph nodes of OT-II mice using CD4⁺ T cell untouched selection kits (Miltenyi Biotec). CD4⁺ T cells were labeled with 2 μmol/L carboxy fluorescein succinimidyl ester (CFSE) (CellTrace kits; Molecular Probes) and added to antigen-pulsed SVCs at a 1:1 ratio in 200 μL of media. After 5 days, T cells were stained for flow cytometry, and CFSE dilution was examined in viable (Live/DeadViolet^-) CD3⁺CD4⁺ lymphocytes (41). T-cell cytokines in the media were measured by ELISA.

CFSE-labeled CD4⁺ OT-II T cells (5 × 10⁶ cells/mouse) were adoptively transferred into lean and obese (14 weeks of HFD) mice by intraperitoneal injection. Imject Ovalbumin (ThermoScientific) or Imject BSA (100 μg/mouse) was administered by intraperitoneal injection 24 h after adoptive transfer. At the indicated times, cells from visceral fat depots, blood, and spleen were analyzed by flow cytometry.

Hybridoma cells secreting MHC I-A/I-E monoclonal antibody were expanded at the University of Michigan Hybridoma Core and purified on protein A/G agarose columns (Pierce). MHC I-A/I-E or rat IgG2b K isotype antibodies (200 μg/mouse) were injected intraperitoneally every 48 h for 18 days.

Glucose tolerance tests were performed on mice fasted (6 h) by intraperitoneal injection of d-glucose (0.7 g/kg body weight). Blood glucose levels (mg/dL) were measured 0, 15, 30, 45, 60, 90, and 120 min after injection using a glucometer (OneTouch Ultra). Plasma insulin levels were measured by ELISA (Crystal Chem).

Data are means ± SEM. Differences between groups were determined using two-tailed Student t tests with GraphPad Prism 5.01 software. P < 0.05 was considered significant.

CD4⁺ ATTs have an activated phenotype in obesity (19,21,28), but direct evidence of T-cell proliferation in fat is lacking. Therefore, we examined CD4⁺ T-cell proliferation within adipose tissue in lean mice and in mice fed HFD by BrdU incorporation (Fig. 1A). HFD increased the number of CD3⁺ CD4⁺ and CD3⁺ CD8⁺ ATTs in epididymal (eWAT) and inguinal (iWAT) white adipose tissue (Supplementary Fig. 1). HFD-induced obesity increased the percentage proliferating BrdU⁺ CD4⁺ ATTs in eWAT but not in iWAT (Fig. 1B). The number of BrdU⁺ CD4⁺ T cells in spleen trended to be higher in response to HFD (Fig. 1B), but the number of BrdU⁺ CD4⁺ T cells in blood, thymus, and lymph nodes was comparable between lean and obese mice (data not shown).

Proliferation of effector-memory (CD44^High CD62L^–), central-memory (CD44^High CD62L⁺), and naïve (CD44^Low CD62L⁺) CD4⁺ ATT subsets was also determined. Consistent with other reports (28), E/M CD4⁺ ATTs were the predominant CD4⁺ ATTs in eWAT (Supplementary Fig. 2). HFD exposure specifically increased E/M CD4⁺ ATT proliferation in fat and did not alter C/M or naïve T-cell proliferation (Fig. 1C). Foxp3⁺ (Treg) and Foxp3^– (Tconv) CD4⁺ T cells in eWAT both incorporated BrdU (Fig. 1D); however, HFD induced an increase in Tconv proliferation but did not alter Treg proliferation (Fig. 1E).

Within eWAT from obese mice, CLS contained BrdU⁺ CD4⁺ cells (Fig. 1F). By contrast, few CD4⁺ T cells were identified in contact with adipocytes in lean fat (data not shown). However, large clusters of BrdU⁺ CD4⁺ T cells were observed on the surface of visceral fat in ND and HFD mice (Fig. 1G and data not shown). These structures resembled FALCs or milky spots in mesenteric and omental fat depots that contain macrophages and lymphoid cells (42,43). In lean eWAT, FALCs contained Foxp3⁺ lymphocytes surrounded by CX3CR1⁺ macrophages (Fig. 1H).

Increased CD4⁺ ATT-cell proliferation in fat suggests an interaction with MHC II⁺ APCs. Macrophage galactose-type C-type lectin (MGL1)⁺ MHC II⁺ ATMs were observed throughout eWAT from lean and obese mice (Fig. 2A–C). MHC II expression in ATMs was concentrated in CLS in obese eWAT (Fig. 2D–E) and in FALCs from lean and obese mice (Fig. 2F and data not shown). FALCs/milky spots in omental fat contained rare populations of MHC II⁺ B220⁺ B cells, but most of the MHC II⁺ cells in FALCs from eWAT were B220^– by immunofluorescence (data not shown).

H2-Ab1 (MHC II) expression increased in eWAT with HFD but was reduced in iWAT and unchanged in the spleens of obese mice (Fig. 2G). The expression of the class II MHC transactivator (CIITA) was also induced in eWAT with HFD (Fig. 2H). Ciita cell-type specific promoters were assessed, and the induction in eWAT was attributed to specific activation of the DC-specific promoter (type I), which was the dominant Ciita promoter expressed in adipose tissue (Fig. 2I). The expression of the B-cell–specific Ciita promoter (type III), but not the IFN-γ inducible promoter (type IV), was modestly induced in eWAT by HFD (Fig. 2I).

SVF from ND and HFD mice were separated into F4/80⁺ CD11b⁺ ATMs and F4/80⁻ CD11b⁻ non-ATM fractions by FACS to further examine H2-Ab1 induction. H2-Ab1 expression was significantly higher in ATMs than in the ATM-depleted SVCs in lean and obese mice, and HFD-induced increase in H2-Ab1 expression was limited to the ATM population (Fig. 2J). Flow cytometry experiments confirmed that F4/80⁺ CD11b⁺ ATMs were the predominant MHC II⁺ population in eWAT (Fig. 2K). In lean mice, 55% of ATMs and less than 10% of non-ATMs expressed MHC II (Fig. 2L). HFD-induced obesity caused a significant increase in the number of MHC II⁺ ATMs (80%) and non-ATMs (Fig. 2L). MHC II surface expression was higher on type 1 CD11c⁺ ATMs than on CD11c⁻ ATMs in HFD mice (Fig. 2M). In sum, these data demonstrate that ATMs are the predominant cell in the SVF capable of communication with T cells via MHC class II–dependent signals in lean and obese mice.

The induction of T cell costimulatory molecules (e.g., CD40, CD80, and CD86) in fat was also assessed. Cd40 and Cd80 were induced by HFD in eWAT (Fig. 3A). Cd40, Cd80, and Cd86 expression was unchanged in the spleen and suppressed in iWAT from obese mice (Fig. 3B and C). F4/80⁺ CD11b⁺ ATMs expressed higher levels of Cd40 than F4/80⁻ CD11b⁻ SVCs in both lean and obese eWAT, and Cd40 was only induced by HFD in ATMs (Fig. 3D).

To examine the mechanisms of costimulatory molecule induction, purified ATMs were cultured in the presence of M1 and M2 macrophage-skewing stimuli. Lipopolysaccharides (LPS) induced Cd40 and Cd80 expression in ATMs, whereas IL-4 had no effect (Fig. 3D and E). IFN-γ alone or combined with LPS enhanced Cd40 but not Cd80 expression in ATMs (Fig. 3D and E).

To examine the capacity of ATMs to phagocytose and process antigens, we performed experiments with OVA as a model antigen. MGL1⁺ ATMs throughout visceral fat and within FALCs rapidly accumulated fluorescently labeled OVA after intraperitoneal injection (Fig. 4A and B). In SVF cultures, the phagocytic capacity of cultured ATMs was heterogeneous (Fig. 4C), and OVA uptake was not an exclusive feature of MHC II⁺ or CD11c⁺ ATMs (Fig. 4D and E). F4/80⁺ ATMs were also capable of efficiently phagocytosing of FITC-labeled ZymosanA and Escherichia coli (Fig. 4F and data not shown).

To determine whether ATMs in visceral fat can proteolytically process antigens, DQ-OVA was injected into mice. As expected, peritoneal macrophages (F4/80⁺ CD11b^High) processed DQ-OVA (Fig. 4I), and ∼20% of F4/80⁺ CD11b⁺ MHC II^High ATMs from eWAT processed DQ-OVA (Fig. 4G). MHC II⁺ F4/80⁻ CD11b⁻ SVCs in eWAT, which contain B cells and nonmyeloid DCs, failed to process DQ-OVA (Fig. 4H). Approximately 4% of MHC II⁻ F4/80⁻ CD11b⁻ SVCs (fibroblasts/preadipocytes) processed detectable amounts of DQ-OVA (Fig. 4H). Antigen processing in fat from obese mice could not be definitively assessed using this approach due to significant autofluorescence from obese SVCs (data not shown).

We next measured the capacity of ATMS to trigger antigen-specific T-cell proliferation by culturing OVA-pulsed ATMs with CD4⁺ T cells from OVA-specific OT-II mice. SVCs from lean mice promoted CD4⁺ T-cell proliferation in an antigen dose-dependent manner (Fig. 5A and B). HFD-induced obesity significantly enhanced the ability of SVCs to induce T-cell proliferation. Antibodies blocking MHC II demonstrated that this effect was MHC II–dependent (Fig. 5C). FACs-purified (F4/80⁺ CD11b⁺) ATMs from obese mice were capable of inducing antigen-specific T-cell proliferation in vitro, but the ATM-depleted SVCs could not (Fig. 5D). SVCs pulsed with full-length OVA protein also induced OT-II CD4⁺ T-cell proliferation (Fig. 5E).

To determine whether SVCs shape T-cell polarization of antigen-specific T cells, cytokines were measured in the cocultures. SVCs from lean and HFD-fed mice both promoted the accumulation of IFN-γ in the media in an antigen dose-dependent manner, but IFN-γ levels were higher when SVCs from obese mice were used as APCs (Fig. 5F). IL-10 and IL-4 production was unaltered (Fig. 5F). Notably, IL-17 production was attenuated in an antigen dose-dependent manner, whereas IL-2 was increased (Fig. 5G). IFN-γ and IL-2 production was dependent on signals from ATMs, because F4/80⁻ CD11b⁻ SVCs were unable to trigger a Th1 skewed T-cell response (Fig. 5H). Overall, these experiments suggest that ATMs from obese mice potently activate and skew naïve T cells toward a Th1 phenotype.

An adoptive transfer approach was used to examine antigen-specific CD4⁺ T-cell proliferation in visceral fat. CFSE-labeled OT-II CD4⁺ T cells preferentially homed to visceral fat depots and localized to FALCs 24 h after adoptive transfer (Supplementary Fig. 3). OVA, but not BSA, induced proliferation of adoptively transferred OT-II CD4⁺ T cells in eWAT, omental, and mesenteric fat depots, as indicated by CFSE dilution and by increased numbers of donor-derived (CD45.2⁺) CD4⁺ T cells recovered from these fat depots (Fig. 6A and B). The total number of CD4⁺ T cells increased in visceral fat depots, but not in spleen, 3 days after OVA exposure (Fig. 6C).

OT-II CD4⁺ T cells were transferred into obese mice to determine the effects of antigen-specific CD4⁺ T-cell activation on adipose tissue inflammation in obese mice. In eWAT, OVA increased the number of CD3⁺ ATTs (expressed as total cells per gram or as a percentage of SVCs) but had no effect on ATM content (Fig. 6D and E). The accumulation of CD4⁺ T cells within fat was attributed not to increases in circulating T cells (Fig. 6F) but, rather, to localized proliferation, because OVA exposure increased the number of CD4⁺ ATTs expressing the cell cycle antigen Ki-67 (Fig. 6G). However, ATT proliferation alone did not alter ATM composition in eWAT (Fig. 6H) and had no significant impact on blood glucose or plasma insulin levels in obese mice (Fig. 6I).

We next tested whether MHC II blockade could modify T-cell inflammation and insulin resistance in mice with established obesity. MHC II immunotherapy (blocking antibody injections every 48 h for 18 days) in mice fed HFD for 14 weeks reduced the total number of CD3⁺ lymphocytes and the number of CD4⁺ ATTs in obese eWAT (Fig. 7A and B). Treg content in eWAT trended toward reduction after MHC II neutralization (Fig. 7C). Notably, there were no differences in CD4⁺ T-cell content in spleen or lymph nodes in obese mice after MHC II neutralization (Fig. 7D). MHC II neutralization did not attenuate the expression of inflammatory genes in eWAT (Tnfa, Il6, Il10, and Ifng; Fig. 7D). Moreover, blood glucose and insulin levels were unaffected (Fig. 7F), and glucose intolerance was similar between IgG and αMHC II–treated HFD mice (Fig. 7G). MHC II neutralization in mice fed HFD for a shorter duration (6–8 weeks) also failed to significantly improve metabolic parameters (Supplementary Fig. 4).

A diverse array of adipose tissue leukocytes is known to impact metabolic disease, yet we have a limited understanding of how leukocytes communicate within fat. Here, we show that ATMs from visceral fat meet the functional definition of APCs. ATMs process and present antigen on MHC II molecules, express T-cell costimulatory molecules, induce antigen-specific CD4⁺ T-cell proliferation, and provide accessory signals to promote the polarization of naïve CD4⁺ T cells into IFN-γ–producing Th1 effector cells. All of these functions were enhanced after dietary obesity due in part to the specific activation of Ciita via a DC/macrophage–specific promoter in ATMs. These data combined with the microscopic evidence of direct ATM–T-cell contact in fat support the concept that ATMs help coordinate an adaptive response to obesity by modulating T-cell activation in fat.

Our data indicate that the capacity for T-cell stimulation is an intrinsic feature of ATMs in lean mice. This provides a system whereby signals early in the course of obesity can be translated into a changed T-cell activation state, at least in part, via the ongoing interaction between CD4⁺ T cells and ATMs. In this way, resident ATMs may function as key sensors of the adipose tissue environment and translate signals from metabolically stressed or damaged adipocytes into an adaptive T-cell response. The potential signals that may feed into and modify this pathway are numerous and include adipokines, fatty acids, cytokines, activators of the NLRP3 inflammasome, and inflammatory CD8⁺ T cells. This model is consistent with the observations that ATT accumulation precedes the appearance of CD11c⁺ ATMs in obese visceral fat (19,21). Our data also indicate that the capacity of ATMs to induce proliferation and IFN-γ production in antigen-specific CD4⁺ T cells is dramatically enhanced by dietary obesity.

Because a significant fraction of ATMs expressing MHC II localize to FALCs that harbor proliferating ATTs, we believe that these atypical structures are important foci for ATM–T-cell interactions in visceral adipose tissue in mice. Notably, interactions between ATMs and ATTs redistribute more broadly with obesity, becoming centered on CLS—regions of active phagocytosis, dense CD11c⁺ ATM accumulation, and cytokine production. Collectively, this suggests that the APC function of ATMs may have duel roles, initially functioning to activate ATT proliferation early in response to obesity and later functioning to retain effector-memory ATTs in adipose tissue as obesity progresses. Technical limitations prevented us from definitively concluding which ATM subtype (M1 or M2) predominates in terms of APC function (data not shown). More studies are needed to define the spatial and temporal dynamics of ATM and ATT interactions in visceral fat at the onset of metabolic stress induced by overnutrition and throughout the progression of obesity.

Our group and others have primarily relied on surface markers to characterize ATMs and putative DCs in fat. Obesity induces the accumulation of myeloid cells (CD11b⁺) that express both macrophage (F4/80) and DC markers (CD11c) in fat, generating a confusing picture of what these cells are functionally. Given the uniqueness of the adipose tissue microenvironment, it is not surprising that myeloid cells in fat will have features that differ from classical lymphoid tissues, such as the spleen. The difficulty in distinguishing between tissue macrophages and DCs is not new to the immunology field, and many have questioned if a distinction between tissue macrophages and DCs is appropriate (44). Our studies have shifted toward clarifying the functional definition of ATMs and how this changes in obesity. A recent study described populations of F4/80⁻ CD11c⁺ B220⁻ adipose tissue DCs that are capable of promoting Th1 and Th17 T-cell responses (45). In our experience, these cells are rare in mice, because >80% of CD11c⁺ cells in visceral fat coexpress CD11b and F4/80. In our in vitro T-cell assays using FACS-sorted cells, F4/80⁻ CD11c⁺ B220⁻ cells were excluded from the F4/80⁺ ATM population but were present in the non-ATM fraction. However, these non-ATMs did not robustly stimulate T-cell proliferation or IFN-γ production in our hands. Although we cannot exclude the possibility that other SVCs may function as APCs, our data strongly imply that F4/80⁺ CD11b⁺ ATMs in visceral fat from mice are robust and efficient APCs that are significantly modified during dietary obesity.

Our initial attempts to improve insulin sensitivity in HFD-fed mice by blocking antigen presentation to CD4⁺ T cells with MHC II–blocking antibodies were unsuccessful in early and late obesity. MHC II neutralization decreased CD4⁺ ATTs specifically in the visceral fat of obese mice, supporting the notion that MHC II–restricted antigen presentation is responsible for activating and retaining CD4⁺ ATTs in fat. However, treating obese mice with MHC II–neutralizing antibodies alone did not improve whole-body glucose homeostasis or alter inflammatory gene expression in fat. These findings are consistent with variable results seen in other studies, and several laboratories have shown that blocking T-cell action alone is not sufficient to improve the metabolic profile in mouse models with established obesity. For example, depletion of CD4⁺ T cells in fat with anti-CD3 antibodies had no effect on glucose tolerance in older obese animals (28), and augmenting T-cell activation in Apoe^−/− mice also had no effect on insulin resistance (46). Mice deficient for CD40 L demonstrated a significant decrease in ATTs, without any improvement in metabolic measures with obesity (47). Likewise, Treg expansion with IL-2 did not lead to significant changes in glucose or insulin tolerance in mice (19). Resolving these observations with the evidence that T cells can impact metabolism when restored in lymphopenic mice will require a more detailed view of the kinetics of T-cell activation and deactivation in fat. One explanation of why MHC II neutralization did not impact glucose metabolism may be due to the limited duration of therapeutic intervention. Two weeks of intervention may not be long enough to fully deactivate IFN-γ–producing CD4⁺ cells in fat. Another explanation may be related to the inability of these treatment modalities to augment the activity of proinflammatory CD11c⁺ ATMs, which are quantitatively the dominant proinflammatory leukocyte population in obese fat.

The identification of an antibody signature in human obesity (24) and our results suggest that protein antigens may trigger an adaptive immune response to obesity. The antigens that give rise to such responses remain elusive and will require further investigation. The limited number of total CD4⁺ and CD8⁺ ATTs in visceral fat (on the order of 1–5 × 10⁴ cells/g) poses a significant obstacle to identifying immunogenic peptide antigens at this time. These challenges also exist in the atherosclerosis field, where T cells have been implicated for over a decade, but only a few causal antigens have been identified (48).

Beyond the identification of the individual protein antigens, the adipose tissue environment itself may significantly modify the ability of antigens to activate T cells. Antigens that are coupled to unsaturated fatty acid–based liposomes can be processed and presented on MHC I and II molecules to activate CD8⁺ and CD4⁺ T cells, respectively; by contrast, antigens coupled to saturated fatty acid–based liposomes are processed and presented via MHC II exclusively (49). In this article, we have chosen to focus on MHC II restricted CD4⁺ T-cell activation based on the current evidence that CD4⁺ T cells have a restricted T-cell receptor repertoire (19,21,28). Although we performed these experiments with non–liposomal-associated peptide antigens, whether putative obesity-induced antigens are liposomally encapsulated is unknown. This may be a potentially important node of regulation of both CD8⁺ and CD4⁺ ATTs.

In summary, we have shown that ATMs in visceral fat are efficient APCs capable of inducing the proliferative expansion of Th1 skewed effector/memory CD4⁺ T cells. Our observations provide a conceptual advancement to our understanding of the complexity of leukocyte networks in adipose tissue. Future studies are necessary to define how these dynamic interactions may contribute to the pathogenesis of metainflammation and metabolic diseases.

This study was supported by National Institutes of Health (NIH) grants DK-090262, DK-078851, DK-087999, and DK-092873 to C.N.L. D.L.M. was supported by NIH National Research Service Award DK-091976. G.M.-S. was supported by NIH Cellular and Molecular Biology Training Grant T32-GM-007315. K.S. was supported by NIH Training Grant T32-HD-008513 and a Pediatric Endocrine Society Research Fellowship Award. This work used services at the University of Michigan and Animal Phenotyping Core, supported by NIH Grant DK-089503, and Chemistry Core(s) at the Michigan Diabetes Research and Training Center funded by P60-DK-020572 from the National Institute of Diabetes and Digestive and Kidney Diseases.

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

D.L.M., K.W.C., and C.N.L. designed and performed experiments, analyzed data, and wrote the manuscript. J.L.D., K.E.O., L.M.G., G.M.-S., and K.S. performed experiments, acquired data, and performed analysis. C.N.L. 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 analysis.

The authors thank Dr. Cheong-Hee Chang, University of Michigan, for the MHC I-A/I-E hybridoma and for providing technical expertise with the design of antigen-specific T-cell assays.