Obesity induces a phenotypic switch in adipose tissue macrophage polarization (original) (raw)
F4/80+CD11b+CD11c+ macrophages accumulate in fat with diet-induced obesity. As the majority of previous studies have used F4/80 expression as a single marker to identify ATMs, we sought to examine ATM heterogeneity in lean and obese mice using additional known macrophage markers. Epididymal fat tissue plays a significant role in overall metabolism in rodents (22). We thus isolated stromal vascular fraction (SVF) cells from epididymal fat pads excised from male mice fed a normal diet (ND) or HFD and analyzed cells by flow cytometry. ATMs from both lean and obese mice coexpressed F4/80 and CD11b, and the majority of the F4/80+CD11b+ cells also expressed Mac-3 (data not shown). However, there was variation in the amount of CD11c expression within the F4/80+CD11b+ ATMs. The SVF from lean mice contained 17.8% F4/80+ cells, consistent with other reports (5). Only a small percentage of these cells coexpressed CD11c (9.3% ± 7.5%; n = 3 pools of mice) (Figure 1A). In contrast, obese mice contained increased F4/80+ cell content within the SVF, with a significant increase in the percentage of F4/80+CD11c+ cells (38.4% ± 15% of F4/80+ cells; P = 0.04 versus lean mice). These cells also expressed CD11b. Similar results were seen in leptin-deficient ob/ob mice, with 74% of the F4/80+ cells coexpressing CD11c (data not shown).
Accumulation of F4/80+CD11c+ ATMs in adipose tissue in obese mice. (A) Analysis of SVF cells for F4/80 and CD11c. Epididymal fat pads from age-matched male C57BL/6 (C57) mice on ND or HFD (n = 3 mice, each condition) were dissected and separated into adipocyte and SVF populations. SVF cells were stained with antibodies against F4/80, CD11c, and isotype controls (open) and analyzed by flow cytometry. Samples were gated for F4/80+ cells and examined for coexpression of CD11c (lower panels). Data from a representative experiment are shown. The percentage of CD11c+ cells within the F4/80+ ATM population is indicated for each condition. (B and C) Quantitation of CD11c+ and CD11c– ATM subpopulations in epididymal fat pads. Flow cytometry was used to assess the percentages of F4/80+CD11c+ and F4/80+CD11c– ATMs in SVF samples from ND- and HFD-fed C57BL/6 mice and HFD-fed CCR2KO mice (n = 3–4 mice per condition). Data are presented as total number of cells per mouse for each ATM subtype (B) and as cell counts normalized to cell number and fat pad weight (C). Data are presented as mean ± SD *P < 0.05 versus ND. (D) Analysis of CD11c expression in ATMs isolated from epididymal fat pads from male CCR2KO mice on a C57BL/6 background on ND or HFD. Cell isolation and flow cytometry were performed as described for A.
The total number of both F4/80+CD11c+ cells (4.52 ± 1.1 × 106 cells/mouse versus 0.196 ± 0.16 × 106 cells/mouse; P = 0.021) and F4/80+CD11c– cells (1.98 ± 0.42 × 106 cells/mouse versus 9.69 ± 8.2 × 106 cells/mouse; P = 0.24) isolated from epididymal fat pads increased in mice fed an HFD compared with lean mice (Figure 1B). Normalizing this data to fat pad weight indicated that there was a 5.5-fold increase in the content of F4/80+CD11c+ cells in adipose tissue from HFD-fed compared with ND mice (Figure 1C). In contrast, there was no significant change in the quantity of F4/80+CD11c– cells in adipose tissue per milligram fat weight between ND and HFD mice. This demonstrates that F4/80+CD11c+ cells are a specific population of ATMs recruited to fat tissue upon HFD exposure.
Previous studies indicated that targeted deletion of the Ccr2 gene results in decreased macrophage content in adipose tissue and improved insulin sensitivity (10). Thus, CCR2KO mice were also examined for CD11c expression as a complementary mouse model. HFD-fed CCR2KO mice demonstrated a reduction in the content of F4/80+CD11c+ cells in the SVF compared with wild-type obese mice, with 14.4% ± 9.2% of the F4/80+ cells coexpressing CD11c (P = 0.08 versus HFD C57BL/6) (Figure 1D). Normalizing for cell number and fat weight indicated that there was a greater than 50% reduction in the level of F4/80+CD11c+ ATMs in the CCR2KO mice (HFD C57BL/6, 2.4 ± 0.46 × 103 cells/mg versus HFD CCR2KO, 0.88 ± 0.98 × 103 cells/mg) (Figure 1C). The content of total F4/80+CD11c+ cells in fat pads from HFD CCR2KO mice was also lower than in obese controls after normalization to cell number (Figure 1B). Similar to what was observed previously (10), adipocyte size was increased in HFD CCR2KO mice compared with control HFD mice (mean 7,933 ± 2,754 μm2 versus 6,704 ± 2,961 μm2, P = 0.041) and HFD CCR2KO mice were protected from hepatic steatosis (data not shown). Contrary to these observations, however, we noted increased fat pad weight in HFD CCR2KO mice (2.83 ± 0.57 g versus control, 1.41 ± 0.25 g; P = 0.011).
Immunofluorescence microscopy confirmed the flow cytometry results; ND SVF cultures contained only F4/80+CD11c– cells (Figure 2A). SVF cells from HFD mice showed a mix of F4/80+CD11c+ and F4/80+CD11c– cells. Immunohistochemistry on epididymal fat pads from obese mice localized CD11c+ cells in macrophage clusters surrounding adipocytes (Figure 2B). The CD11c antibody stained a subset of cells in these crownlike structures, while F4/80 stained nearly all of the cells in these clusters. Scattered individual CD11c+ ATMs were observed surrounding adipocytes in what may have been early ATM clusters (Figure 2C). Rare CD11c+ cells were seen in adipose tissue sections from lean mice.
CD11c+ ATMs in SVF cultures and in adipose tissue from obese mice. (A) Identification of F4/80+ and CD11c+ ATMs by immunofluorescence microscopy. Epididymal fat pads from ND and HFD mice were separated into adipocyte and SVF fractions. SVF cells were plated onto glass coverslips and cultured overnight prior to fixation. Cells were stained with antibodies against F4/80 (left) and CD11c (middle) and imaged by confocal microscopy to identify surface markers to confirm the presence of CD11c+ cells only in the SVF from HFD-fed animals. Similar results were obtained for 3 independent sets of cultures. (B) Immunohistochemical localization of CD11c+ in adipose tissue. Consecutive sections from epididymal fat pads from obese C57BL/6 mice were stained with anti-F4/80 (left panels) and anti-CD11c antibodies (right panels), followed by colorimetric detection (brown). Sections were counterstained with hematoxylin (blue) and images taken at low (×200) and high magnification (×1,000). (C) In obese mice, CD11c+ cells were also detected surrounding normal-appearing adipocytes in the absence of crownlike macrophage clusters (×1,000 magnification).
Increased inflammatory gene expression in F4/80+CD11c+ ATMs. Because CD11c+ cells appear to accumulate in adipose tissue after high-fat feeding, we assessed expression of inflammatory genes in F4/80+CD11c+ ATMs compared with F4/80+CD11c– ATMs from obese mice. Real-time RT-PCR showed appropriate sorting of the CD11c populations based on expression of integrin, alpha X (Itgax, encoding CD11c) (Figure 3). F4/80+CD11c+ cells overexpressed the inflammatory genes Il6 (IL-6) and Nos2 (iNOS) compared with F4/80+CD11c– ATMs. Tnfa (TNF-α) was not differentially expressed between the cell types, while Apoe (apoE) was downregulated in the F4/80+CD11c+ population relative to F4/80+CD11c– cells.
Increased inflammatory gene expression in F4/80+CD11c+ ATMs. SVF cells were isolated from HFD-fed male C57BL/6 mice (n = 3) and stained for F4/80 and CD11c. F4/80+CD11c+ and F4/80+CD11c– cells were isolated by FACS and total RNA isolated. Expression of Itgax (CD11c), Tnfa (TNF-α), Il6 (IL-6), Nos2 (iNOS), and Apoe (apoE) was analyzed by real-time RT-PCR in F4/80+CD11c+ (white bars) and F4/80+CD11c– (black bars) ATMs. Data are expressed as mean ± SD. *P < 0.05.
ATMs from lean mice express markers of alternatively activated macrophages. Given the low level of inflammatory cytokine expression in adipose tissue from lean mice, we examined the ATMs derived from lean mice for known markers of alternatively activated M2 macrophages (15, 23). ATMs from lean animals showed increased expression of M2-specific genes Il10 (IL-10), Arg1 (arginase I), Mrc2 (mannose receptor, C type 2), Ym1/Chi3l3 (Ym1/chitinase 3–like 3), and Mgl1/2 (macrophage galactose _N_-acetyl-galactosamine–specific lectins 1 and 2) compared with the ATMs isolated from HFD mice (Figure 4A). In contrast, expression of proinflammatory genes such as Tnfa and Nos2 was considerably lower in ND ATMs compared with ATMs from obese animals (Figure 4B).
Increased expression of markers of alternatively activated (M2) macrophages in ATMs from lean ND-fed mice. (A and B) Gene expression in ATMs from ND and HFD mice. F4/80+CD11b+ ATMs were isolated from ND C57BL/6 (white bars), HFD C57BL/6 (black bars), and HFD CCR2KO mice (gray bars) (n = 3 pools of mice for each) and analyzed by real-time RT-PCR for expression of M2 macrophage–specific genes (A) and proinflammatory genes (B). Data are expressed as mean ± SD. *P < 0.05. (C and D) SVF was isolated from ND (white bars) and HFD (black bars) mice (n = 2–3 mice per condition) and analyzed by real-time RT-PCR for expression of M2 macrophage markers (C) and proinflammatory genes (D). (E) Ym1 protein expression in the SVF. Lysates from SVF isolated from ND and HFD mice were immunoblotted for Ym1 (left). CD11b+ ATMs were separated from CD11b– cells in the SVF and lysates prepared for immunoblotting, which demonstrated Ym1 expression in the macrophage fraction (right). Macrophage marker CD68 controlled for the purification protocol. Similar results were obtained in a duplicate experiment. (F) Arginase activity in adipose tissue from ND and HFD mice. Epididymal fat pads from ND- (white bars) and HFD-fed (black bars) mice were separated into adipocyte and SVF fractions and lysates prepared. Arginase activity was assessed by an assay of urea production from arginine substrate and was normalized to protein concentration. Reactions were performed in triplicate. Data are expressed as mean ± SD. *P < 0.05. Similar results were obtained for 3 separate sets of mice.
As macrophages make up only a portion of the stromal cells in adipose tissue, we examined expression of some of these genes in unsorted SVF isolates. Similar patterns of gene expression were seen in RNA prepared from SVF cells from lean and obese mice (Figure 4, C and D), with increased Ym1 and Il10 expression in ND mice and increased Tnfa and Nos2 expression in HFD mice.
Since inflammatory macrophage recruitment into fat tissue appears to be partially dependent on CCR2 (10), we hypothesized that ATMs in obese CCR2KO mice might retain properties of M2-polarized macrophages and thus evaluated gene expression in these cells. Indeed, ATMs isolated from obese CCR2KO mice expressed alternatively activated macrophage markers at levels that were similar to those seen in ATMs from lean mice. Furthermore, these were significantly elevated compared with levels in ATMs from obese mice (Figure 4A).
We confirmed gene expression data by examining protein levels in the ATMs under study. Consistent with the RNA expression data, Ym1 was increased in SVF samples from lean compared with obese mice (Figure 4E). Additionally, purification of CD11b+ ATMs from SVF samples indicated that Ym1 is predominantly expressed in ATMs compared with non-macrophage CD11b– cells.
Another characteristic of the alternative macrophage activation state is increased arginase activity (15). Arginase activity was assessed in SVF and adipocyte samples from lean and obese mice. While the activity of this enzyme did not differ between the isolated adipocytes from lean and obese mice, arginase activity was significantly diminished in the SVF fraction after HFD exposure (Figure 4F). This indicates that the arginase/iNOS balance in ATMs from lean animals is skewed toward an immunosuppressive state that changes to a proinflammatory state with obesity (elevated iNOS and repressed arginase activity).
IL-10 protects adipocytes from the deleterious effects of TNF-α. The observation that the expression of Il10 is higher in the SVF and ATMs from lean compared with obese mice led us to examine the effects of the cytokine on adipocyte function. To explore the cellular effects of IL-10, we first examined the expression of its receptor. The IL-10 receptor α (IL-10Rα) was detected in adipose tissue by immunoblotting (Figure 5A). Separation of the SVF from adipocytes in lean and obese animals indicated that the IL-10 receptor was expressed primarily in adipocytes, with minimal expression in the SVF (Figure 5B). No significant differences were seen in IL-10Rα expression between lean and obese mice.
IL-10 signaling in adipocytes. (A) Expression of IL-10 receptor in adipose tissue. Immunoblots of lysates from adipose tissue (Ad), lung (Lu), and spleen (Sp) from mice probed with IL-10 receptor α (IL-10Rα) antibodies. (B) Expression of IL-10 receptor in adipocytes and not the SVF from adipose tissue. Epididymal fat pads from ND- and HFD-fed mice were collected and separated into SVF and adipocyte fractions. Lysates were prepared and immunoblotted with anti–IL-10 receptor antibodies and loading controls. (C) IL-10 treatment of adipocytes activates STAT3 and Akt. Differentiated 3T3-L1 adipocytes were stimulated with IL-10 (20 ng/ml) for the indicated times and lysates prepared. Immunoblots were probed with antibodies against phosphoY705-STAT3 and phosphoS473-Akt and STAT3 and Akt as loading controls. Experiments were repeated twice, and results of a representative experiment are shown.
We next examined IL-10 signaling in differentiated 3T3-L1 adipocytes. STAT3 phosphorylation by the IL-10 receptor has been shown to be required for its antiinflammatory effects (24). Treatment of adipocytes with IL-10 led to the tyrosine phosphorylation of STAT3 within 5 minutes. This effect peaked at 10 minutes and declined thereafter (Figure 5C). As IL-10 as been shown to activate the PI3K pathway via insulin receptor substrate (IRS) proteins (25), we examined phosphorylation of Akt in response to IL-10. IL-10 treatment of adipocytes produced the rapid serine phosphorylation of Akt (Figure 5C).
IL-10 has a variety of antiinflammatory effects in macrophages and other cell types that includes the antagonism of the actions of TNF-α (26). To examine this in adipocytes, we first analyzed the effect of IL-10 on the secretion of the chemokine MCP-1, thought to be a major attractant for macrophages in states of obesity (9, 11, 27). IL-10 treatment for 16 hours decreased MCP-1 secretion from adipocytes (Figure 6A). As reports suggest that IL-10 may mediate its antiinflammatory effects by antagonizing activation of NF-κB–mediated gene transcription (28), we examined the effects of IL-10 on TNF-α–stimulated NF-κB activity using an NF-κB–responsive luciferase reporter gene. While TNF-α induced reporter gene expression, IL-10 had no effect upon this activation (data not shown). This is consistent with other reports that show that IL-10 can alter the transcriptional rate of inflammatory genes independent of NF-κB activation (29).
IL-10 prevents the effects of TNF-α on adipocytes. (A) IL-10 decreases MCP-1 secretion by adipocytes. 3T3-L1 adipocytes were treated with media with or without IL-10 (20 ng/ml) for 16 hours. Conditioned media was then removed and assayed for MCP-1 levels by ELISA. n = 5 independent samples per condition. Data are expressed as mean ± SD. (B) IL-10 protects adipocytes from TNF-α–induced downregulation of insulin receptor and glucose transporter 4 (GLUT4) expression. 3T3-L1 cells were treated with or without IL-10 for 24 hours prior to treatment with or without TNF-α (17 ng/ml for 6 hours). Lysates were prepared and immunoblots probed with antibodies against the insulin receptor (IR) and GLUT4. (C) IL-10 maintains IRS levels despite treatment with TNF-α. Adipocytes were treated as described for B and lysates examined for IRS1 tyrosine phosphorylation induced by insulin (INS; 100 nM for 5 minutes) after immunoprecipitation of IRS1. IRS serine phosphorylation at Ser307 was evaluated using specific antibodies.
We next examined the effects of IL-10 on insulin signal transduction in adipocytes in concert with TNF-α. Treatment of cells with TNF-α for 6 hours dramatically reduced the levels of glucose transporter 4 (GLUT4) protein, with a smaller reduction in the level of insulin receptor expression, similar to what has been reported previously (30). IL-10 pretreatment prevented the TNF-α–induced downregulation of both proteins (Figure 6B). Additionally, insulin-stimulated IRS1 tyrosine phosphorylation was decreased with TNF-α treatment. This inhibitory effect was prevented by pretreatment with IL-10 (Figure 6C). IL-10 treatment produced no significant changes in TNF-α–stimulated IRS1 phosphorylation at Ser307, suggesting that the antagonism of TNF-α action by IL-10 occurs either downstream of or in parallel to reduced tyrosine phosphorylation of IRS1.
To determine whether IL-10 protects adipocytes from the physiological effects of TNF-α, we examined insulin-stimulated glucose uptake in 3T3-L1 adipocytes with and without IL-10 pretreatment. Treatment of cells with TNF-α for 3 hours produced a significant, 30%–40% inhibition of insulin-stimulated glucose uptake. Pretreatment of the cells with IL-10 for 24 hours blocked these acute inhibitory effects of TNF-α (Figure 7A). Prolonged (72-hour), low-dose TNF-α treatment increased basal glucose uptake but had no effect on insulin-stimulated levels, consistent with a dramatic upregulation of the glucose transporter GLUT1 under these conditions (31). IL-10 blocked these effects of TNF-α and maintained cells in an insulin-sensitive state (Figure 7B). Moreover, insulin-stimulated glucose uptake was dramatically enhanced in cells chronically treated with IL-10 compared with untreated cells.
IL-10 prevents the effects of TNF-α on blocking insulin-stimulated glucose uptake in adipocytes. (A) Pretreatment of adipocytes with IL-10 blocks TNF-α effects on glucose uptake. 3T3-L1 adipocytes were treated with IL-10 (20 ng/ml) for 24 hours prior to treatment with TNF-α (17 ng/ml) for 3 hours. 2-Deoxyglucose (2-DG) uptake was assessed after 30 minutes without (white bars) or with (black bars) insulin (100 nM) stimulation. Data are expressed as mean ± SD of triplicate experiments repeated 3 times. *P < 0.05. (B) Insulin-stimulated glucose uptake in adipocytes chronically treated with IL-10 and TNF-α. Differentiated 3T3-L1 adipocytes were treated with low-dose TNF-α (3 ng/ml) for 72 hours in the presence or absence of IL-10 (20 ng/ml). After insulin stimulation (black bars), 2-DG uptake (upper panel) and fold change in glucose uptake (lower panel) were assessed, and the results demonstrated that IL-10 blocks the effects of TNF-α. Data are expressed as mean ± SD of triplicate experiments repeated 2 times. *P < 0.05.