Weight loss and lipolysis promote a dynamic immune response in murine adipose tissue (original) (raw)
Weight loss induces a transient accumulation of ATMs. To understand how the immune system — and specifically ATMs — responds to weight loss, we characterized the metabolic and inflammatory phenotypes of obese mice that were subjected to moderate caloric restriction. Nine-week-old male C57BL/6J mice were fed a high-fat diet (60% of the calories derived from fat; Supplemental Table 1; supplemental material available online with this article; doi:10.1172/JCI42845DS1) until they reached 40 grams of body mass. These animals were then subjected to caloric restriction (they received 70% of their ad libitum food intake) to induce gradual weight loss. Mice were sacrificed and tissues collected after 0, 3, 7, 14, 21, 42, and 60 days of caloric restriction. A group of age-matched lean-chow diet–fed mice was studied as a control (Supplemental Figure 1A). At the start of caloric restriction, all high-fat diet–fed mice had similar body composition, fasting blood glucose, and serum insulin concentrations (Supplemental Table 2). We rotated the assignment of mice to different caloric restriction groups, ensuring that the mean age of each group at sacrifice was not significantly different (Supplemental Figure 2A). As expected, this protocol of moderate caloric restriction induced a gradual reduction in weight and fat mass without affecting lean mass significantly (Supplemental Figure 1, B and C, and Supplemental Figure 2B). Total fat mass, as measured by NMR spectroscopy, was measurably reduced by day 14. Mice continued losing weight throughout the intervention and at the end of the caloric restriction they had lost 27.9% of their initial body mass (40.8 ± 0.3 g vs. 29.4 ± 3.8 g; P < 0.001). All adipose depots decreased in mass at a similar rate (Supplemental Figure 1D), and the decrease in fat mass was due to a decrease in adipocyte size and not to a decrease in adipocyte number (Supplemental Figure 1, C–F). Decreases in adipocyte size were accompanied, as expected, by a reduction in leptin gene expression and leptin serum concentration (Supplemental Figure 3, A and B).
During weight gain and in cross-sectional studies of weight-stable humans, there is a positive, nearly linear relationship between adiposity and markers of local and systemic inflammation (13). Following weight loss of approximately 17% in patients 3 months after bariatric surgery, ATM content and adipose tissue inflammation are reduced and insulin sensitivity improved (32, 33). However, there have been few studies of the relationship between adiposity and measures of inflammation during dynamic weight loss. A recent study by Langin and colleagues suggests that during early weight loss in humans, the expression of inflammatory genes is not decreased (34).
To provide greater temporal resolution of the relationship between adiposity and metabolic and inflammatory phenotypes during weight loss, we measured fasting blood glucose, serum insulin, and adipose tissue expression of inflammatory genes in our cohort of mice during 2 months of continuous weight loss. Fasting blood glucose and serum insulin concentrations remained elevated during the first week of weight loss but thereafter significantly decreased, concomitantly with the reduction in percentage of body fat (Supplemental Figure 4). In contrast, the reduction in inflammatory gene expression in perigonadal adipose tissue was not uniform, and several classes of genes were identified based on their expression pattern during weight loss. The expression of some inflammatory genes such as Saa3 decreased early during weight loss, preceding the improvement in glucose homeostasis, whereas the gene expression of other prototypical M1 inflammatory genes, including Tnf (Tnfa) remained elevated during the entire period when the animals were in negative energy balance (Supplemental Figure 5). Circulating inflammatory proteins also fell into several classes in response to negative energy balance, with concentrations of some inflammatory molecules, e.g., resistin, falling early during weight loss and before metabolic improvement, while the concentrations of others, e.g. PAI-1, remained unchanged throughout caloric restriction and weight loss (Supplemental Figure 3, C and D).
In contrast to the other classes of genes, the adipose tissue expression of macrophage/myeloid cell–specific genes Emr1 (F4/80), Cd68, and Csf1r increased after 3 days of caloric restriction (Figure 1A). By 60 days of weight loss, however, the expression of Emr1 and Cd68 was decreased to levels below those present prior to the start of weight loss. The late reduction in macrophage-specific gene expression was consistent with previous studies that had examined the effects of long-term weight loss. To determine whether the initial increase and ultimate reduction in macrophage/myeloid-specific gene expression was due to alterations of gene expression or to a change in macrophage number, we performed immunohistochemistry using an antibody that recognizes the macrophage antigen F4/80 (EMR1). Consistent with the gene expression studies, macrophage number in perigonadal adipose tissue increased during the first week of weight loss. Three days after the beginning of caloric restriction, perigonadal adipose tissue had 47% more macrophages than adipose tissue from ad libitum–fed control mice (percentage of macrophages per total cells: 38.6% ± 4.1% vs. 26.3% ± 7.4%; P < 0.01; Figure 1, B and C). We and others have previously shown that there is a strong positive correlation between adiposity and ATM content (2, 18). However, during weight loss, this relationship is lost. Instead, during an initial period of weight loss (days 0–7 in our study), there are more ATMs found in leaner mice (Figure 1D), whereas later during weight loss, the more typical positive correlation is found (Figure 1D). Identical relationships were found when ATM content was plotted versus adipocyte size (data not shown). ATMs, and specifically CD11c+ ATMs and CD11c+ crown-like structures (CLS), are most tightly associated with adipose tissue inflammation and systemic insulin resistance. Almost all CD11c+ cells in the perigonadal adipose tissue of mice during early weight loss were also F4/80+ (CD11c+ ATMs) (Supplemental Figure 6). Although the total number of ATMs increased during early weight loss, the number of CD11c+ ATMs and CLS did not increase (Supplemental Figure 7), consistent with accumulation of a CD11c– population of ATM during early weight loss and a lack of increase in markers of inflammation or an impairment in insulin resistance during this same period (Supplemental Figures 4 and 5).
ATM content increases, then decreases during weight loss. (A) Expression of genes encoding myeloid-macrophage proteins in perigonadal adipose tissue. Black bars represent high-fat diet–induced obese mice that underwent caloric restriction for different time intervals. White bars represent control lean mice that were fed a chow diet (CD) and did not undergo caloric restriction. n = 5–6 mice/group. (B) Immunohistochemical staining of F4/80-expressing (EMR1) macrophages in perigonadal adipose tissue sections from mice during weight loss following indicated number of days of caloric restriction. Arrows indicate ATMs. Scale bars: 50 μm. (C) Macrophages as a percentage of all cells in perigonadal adipose tissue. n = 5–6 mice/group. (D) Relationship between macrophage content and body weight in mice during the first 7 days of weight loss (left panel) and during days 14–60 of weight loss (right panel). The square values of the Pearson’s correlation coefficients are shown. Each data point represents the % of macrophages in murine perigonadal adipose tissue at different body weights during caloric restriction. (E) Immunohistochemical staining of F4/80-expressing macrophages (EMR1) in subcutaneous adipose tissue sections. Scale bars: 50 μm. (F) Macrophages as a percentage of all cells in subcutaneous adipose tissue from mice during weight loss. n = 5–6 mice/group. All data are represented as mean ± SD. *P < 0.05; **P < 0.01; ***P < 0.001, versus day 0.
The early initial increase in ATM content was not unique to perigonadal adipose tissue. Consistent with previous findings in high-fat diet–induced obese mice, in subcutaneous adipose tissue, the ATM content was lower than in perigonadal adipose tissue (2, 32). However, after 3 days of caloric restriction, the ATM content doubled in subcutaneous depots (macrophages as a percentage of all cells: 10.0% ± 3.1% vs. 20.2% ± 7.4%; P < 0.05; Figure 1, E and F). In both perigonadal and subcutaneous depots, ATM numbers decreased progressively after 3 days of negative energy balance so that after 42 days of caloric restriction, ATM content was significantly lower than in adipose tissue from high-fat–fed mice that had never been calorically restricted (Figure 1, B, C, E, and F).
During weight gain, it has been suggested that ATM accumulation is driven by adipocyte necrosis, tissue remodeling, or microhypoxia. The initial increase in ATMs in response to weight loss was not associated with a reduction in adipocyte number, as might have been expected, if adipocyte necrosis was driving the accumulation of ATMs (Supplemental Figure 1F). Nor did the peak in ATMs coincide with an upregulation of a transcriptional program of adipose tissue remodeling (Supplemental Figure 8).
Measures of lipolysis correlate with ATM content. The initial increase in ATMs was associated with changes in circulating FFA concentrations and measures of lipolysis. Basal lipolysis in adipose tissue is the release of FFA from adipocytes, which occurs in the absence of negative energy balance. Basal lipolysis is increased in adipose tissue of obese individuals and correlates positively with adipocyte size (25). Demand lipolysis is the hormonally and autonomically driven release of FFA from adipocyte triglycerides that is activated by negative energy balance, when FFA are mobilized from adipose tissue for systemic use as substrates (25). A model in which lipolysis regulates ATM accumulation is consistent with previous observations of weight-stable or weight-gaining individuals: obese mice with large adipocytes have higher basal adipose tissue lipolysis and greater ATM content than lean animals. During early weight loss when adipocyte size has not changed significantly, basal lipolysis remains high and demand lipolysis increases, and thus we predict there is a net increase in total lipolysis. But as adipose tissue mass and adipocyte size decrease, basal lipolysis is also reduced and the net efflux of lipids from adipose tissue decreases. Serum FFA concentrations correlate with total rates of lipolysis and fatty acid fluxes in adipose tissue (25, 35). Therefore, if lipolysis does have a role in the accumulation of ATMs, we predicted that serum FFA would correlate with ATM content in our calorically restricted mice. Indeed, serum FFA concentrations were higher on day 3 of caloric restriction, before significant weight loss, and coincident with the peak in macrophage content (Figure 2A). Consistent with our hypothesis, a positive correlation existed between serum FFA concentrations and the percentage of ATMs in perigonadal adipose tissue throughout the duration of caloric restriction (Figure 2B).
Measures of lipolysis correlate with ATM content. (A) Serum concentrations of FFA during weight loss induced by caloric restriction. Black bars represent high-fat diet–induced obese mice that underwent caloric restriction for different time intervals. White bar represents control lean mice that were fed a chow diet and did not undergo caloric restriction. n = 5–6 mice/group. (B) Correlation of macrophage content (% macrophages) and serum FFA concentration in mice during weight loss; the square value of the Pearson’s correlation coefficient is shown. n = 5–6 mice/group. Each data point represents the % of macrophages in murine perigonadal adipose tissue at different serum FFA concentrations. (C) Perigonadal adipose tissue expression of the gene encoding the lipase ATGL in mice during weight loss induced by caloric restriction. n = 5–6 mice/group. (D) FFA release from explants of perigonadal adipose tissue incubated under basal conditions. Explants were isolated from high-fat diet–induced obese mice that were ad libitum fed or underwent caloric restriction for 3 or 42 days. (E) Glycerol release from explants of perigonadal adipose tissue incubated under basal conditions. Explants were isolated from high-fat diet–induced obese mice that were ad libitum fed or underwent caloric restriction for 3 or 42 days. All data are represented as mean ± SD. *P < 0.05, versus day 0.
In adipose tissue, the rate-limiting step in both basal and demand lipolysis is regulated by the enzyme encoded by Atgl/Pnpla2. The expression of Atgl/Pnpla2 is regulated by nutritional status and is closely correlated with rates of adipose tissue lipolysis (36). Consistent with there being a peak of adipose tissue lipolysis on day 3 of caloric restriction, the adipose tissue expression of Atgl/Pnpla2 increased on day 3 of caloric restriction and returned to baseline by day 42 (Figure 2C). In contrast, the expression of hormone-sensitive lipase encoded by Hsl/Lipe is not correlated with nutritional status. Hsl/Lipe mRNA levels are downregulated during acute fasting and increase only after prolonged food deprivation (36). Consistent with these data, we did not observe any changes in Hsl/Lipe levels during caloric restriction (Supplemental Figure 9A).
Circulating FFA concentrations and Atgl/Pnpla2 expression provided indirect measures of adipose tissue lipolysis. To directly measure lipolysis in adipose tissue during caloric restriction, the rates of release of nonesterified FFA and glycerol were measured in perigonadal adipose tissue from mice during caloric restriction. Consistent with our indirect measures, lipolysis was increased in adipose tissue from mice following 3 days of caloric restriction compared with adipose tissue from ad libitum–fed mice (Figure 2, D and E). FFA and glycerol release were reduced after 42 days. These data demonstrate a positive correlation between adipose tissue lipolysis and ATM content. To determine directly whether increasing or decreasing lipolysis alters ATM accumulation, we performed a series of dietary, pharmacological, and genetic manipulations.
Lipolysis induces ATM accumulation. If lipolysis drives the accumulation of ATMs in adipose tissue, then fasting, which rapidly increases adipose tissue hydrolysis of triglycerides, should also increase ATM content. Perigonadal adipose tissue was collected from high-fat–fed obese C57BL/6J mice that were either fasted for 24 hours or fed ad libitum. Fasting induced an increase in serum FFA concentration (Figure 3A) and led to a rapid accumulation of ATMs. Compared with adipose tissue from ad libitum–fed mice, adipose tissue from fasted mice contained 65% more ATMs (percentage of macrophages per total cells: 22.9% ± 6% vs. 37.9% ± 3.5%; P < 0.01; Figure 3, B–D). Fasting-induced ATM accumulation was not limited to obese mice. In lean mice, the expression of macrophage-specific genes, Emr1 and Csf1r, was increased by 3- and 4-fold respectively after a 24-hour fast (Figure 3E). Consistent with our observations that caloric restriction did not induce an accumulation of CD11c+ ATMs, the expression of Itgax (Cd11c) was unchanged by fasting (Figure 3E).
Induction of lipolysis increases macrophage content in adipose tissue. (A) Serum concentrations of FFA in high-fat diet–induced obese ad libitum–fed and 24 hour–fasted mice. n = 5–6 mice/group. **P < 0.01, versus ad libitum fed (Ad lib fed). (B and C) Immunohistochemical staining of F4/80-expressing (EMR1) macrophages in perigonadal adipose tissue sections from high-fat diet–induced obese ad libitum fed (B) and 24 hour-fasted mice (C). Arrows indicate ATMs. Scale bars: 50 μm. (D) Macrophages as percentage of all cells in perigonadal adipose tissue from high-fat diet–induced obese ad libitum–fed and 24 hour–fasted mice. n = 5–6 mice/group. **P < 0.01, versus ad libitum fed. (E) Expression of genes encoding myeloid-macrophage–specific proteins in lean ad libitum–fed and 24 hour–fasted mice. n = 5–6 mice/group. *P < 0.05, versus ad libitum fed. (F) Protocol for pharmacologically induced adipocyte lipolysis through β3-adrenergic agonist (CL316,243) in lean mice. (G–I) Immunohistochemical staining of F4/80-expressing (EMR1) macrophages in perigonadal adipose tissue sections from lean mice treated with vehicle (G) or with CL316,243 (H and I). Multinucleated giant cells containing lipid droplets are apparent in some sections (I). Arrows indicate ATMs. Scale bars: 50 μm. (J) Macrophages as a percentage of all cells in perigonadal adipose tissue from vehicle- and CL316,243-treated mice. n = 5 mice/group. ***P < 0.001, versus vehicle. All data are represented as mean ± SD.
Activation of the β3-adrenergic receptor (Adrb3), which in mice is almost exclusively expressed by adipocytes, increases adipocyte lipolysis. Granneman and colleagues had previously noted that β3-adrenergic stimulation leads to the accumulation of myeloid-appearing cells in adipose tissue (37). To determine whether pharmacological activation of lipolysis would increase ATM accumulation, we injected lean C57BL/6J mice (body weight = 24.5 ± 0.8 g) with the β3-adrenergic agonist CL316,243 twice, 4 hours apart. Adipose tissue depots were collected 14 hours after the final injection (Figure 3F). Compared with control-treated mice, pharmacological induction of lipolysis rapidly increased ATM content in perigonadal adipose tissue more than 5-fold, to levels typical of obese mice (Figure 3, G–J). Macrophages were seen both isolated and in clusters (Figure 3, G–I). Thus, among the hormonal signals activated by fasting, β3-adrenergic induction of adipocyte lipolysis is sufficient to induce rapid macrophage recruitment. Similar to our observations in early weight loss and fasting, the β3-adrenergic–induced recruited macrophages do not increase the inflammatory phenotype of adipose tissue (Supplemental Figure 9E)
Reducing lipolysis during weight loss or fasting decreases ATM accumulation. If our hypothesis is correct, then manipulations that inhibit lipolysis during early weight loss or fasting should reduce or prevent ATM accumulation. While diets high in fats increase dietary lipids, they are also ketogenic during negative energy balance. Compared with high-fat diets during negative energy balance, diets high in carbohydrate content increase circulating insulin/glucagon ratio and reduce lipid mobilization from and lipolysis rates in adipose tissue (38). Therefore, we repeated the caloric restriction intervention, adding a weight-matched group of mice that were calorically restricted on an isocaloric high-carbohydrate diet (Supplemental Table 1). Mice were maintained on caloric restriction for 3 days, and mice in both groups were equally restricted (30% fewer calories than their ad libitum consumption). At the end of caloric restriction, there was no difference in weight between the high-carbohydrate and high-fat diet caloric–restricted groups. However, as expected, the serum FFA concentration of mice calorically restricted on a high-carbohydrate diet was 36% lower than the mice calorically restricted on a high-fat diet (0.38 ± 0.11 vs. 0.59 ± 0.07 mmol/l; P < 0.05) (Figure 4A). Consistent with our hypothesis, during weight loss, ATM content was 35% lower (31.1% ± 4.1% vs. 20.1% ± 5.2%; P < 0.05) in perigonadal adipose tissue in mice fed a high-carbohydrate diet compared with those fed a high-fat diet (Figure 4, B and C).
Lipolysis inhibition through dietary manipulation limits ATM accumulation during early weight loss. A caloric restriction protocol was used to induce weight loss with lower rates of lipolysis compared with caloric restriction of mice on a high-fat diet. High-fat diet–induced obese mice were fed 70% of their ad libitum caloric intake for 3 days in the form of either a diet high in carbohydrate or fat content. (A) Serum FFA in mice during weight loss induced by caloric restriction on a diet high in either fat or carbohydrate content. n = 5–6 mice/group. (B) Perigonadal adipose tissue sections from mice during weight loss induced by a diet high in fat (left panel) or high in carbohydrate content (right panel). Arrows indicate ATMs. Scale bars: 50 μm. (C) Macrophages as a percentage of all cells in perigonadal adipose tissue from mice during weight loss induced by caloric restriction on a diet high in either fat or carbohydrate content. n = 5–6 mice/group. *P < 0.05, versus calorie restriction on HFD. All data are represented as mean ± SD.
Adipose triglyceride lipase (also known as desnutrin or patatin-like phospholipase domain–containing protein 2) (Atgl/Pnpla2) regulates both basal and demand lipolysis in adipose tissue. Animals deficient in ATGL/PNPLA2 are severely impaired in their ability to mobilize FFAs, have very low basal lipolysis, and are unable to mount demand lipolysis in response to fasting, despite having an intact hormonal and autonomic response to fasting (36). To provide genetic evidence that lipolysis is a critical determinant of ATM content, we studied the effects of fasting in _Atgl/Pnpla2_–/– mice. Consistent with our hypothesis, we found that ad libitum_–_fed Atgl/Pnpla2–/– mice have fewer than 3% ATMs and that following a fast, there was no increase in ATMs (Figure 5, A and B) or macrophage-specific gene expression (Figure 5C). These data argue that the lipase ATGL/PNPLA2 is required for ATM recruitment and accumulation in adipose tissue.
ATGL/PNPLA2 deficiency limits ATM accumulation during fasting. (A) Immunohistochemical staining of F4/80-expressing (EMR1) macrophages in perigonadal adipose tissue sections from Atgl+/+ (top panel) and Atgl–/– (lower panel) mice that were either ad libitum fed (left) or fasted (right). Arrows indicate ATMs. Scale bars: 100 μm. (B) Macrophages as a percentage of all cells in lean ad libitum–fed Atgl+/+ and fasted Atgl–/– mice. n = 4–5 mice/group. ***P < 0.001, versus ad libitum fed. (C) Expression of macrophage-specific genes in perigonadal adipose tissue of lean ad libitum–fed and fasted Atgl–/– mice. n = 4–5 mice/group. All data are represented as mean ± SD.
Adipose tissue lipolysis induces macrophage migration. The rate of mitosis in adipose tissue of fasted mice was very low (<2%) and not different from the rate of mitosis in adipose tissue from ad libitum–fed mice (Supplemental Figure 9, B and C), suggesting that lipolysis-dependent increase in ATMs was not due to proliferation but a consequence of myeloid cell recruitment. To determine whether lipolysis increases the release of adipose tissue chemoattractants for macrophages, we performed a migration assay with adipose tissue explants from fasted and ad libitum–fed mice. Perigonadal adipose tissue explants were collected from lean C57BL/6J mice that were either ad libitum fed or fasted for 24 hours. Explants from ad libitum–fed mice were incubated under basal conditions or with the addition of isoproterenol to induce lipolysis. As expected, compared with adipose tissue isolated from ad libitum–fed mice, adipose tissue from fasted mice or adipose tissue treated with isoproterenol increased lipolysis as evidenced by an increase in the release of FFA (Figure 6A). With this increase in lipolysis, there was a proportional increase in the chemotactic activity of adipose tissue toward bone marrow–derived macrophages (BMDMs). This increase in adipose tissue chemoattractant activity was comparable to that induced by MCP-1/CCL2 (Figure 6B). We also found that in contrast to fasting wild-type mice, fasting CCR2-deficient mice did not increase the macrophage-specific gene expression (Supplemental Figure 9D). However, there was no increase in adipose tissue expression of known CCR2 ligands during a fast (data not shown). These data are consistent with the recruitment of ATMs during a fast being a multi-step process in which accumulation of ATM precursors in the circulation is CCR2 dependent but transit into fat depot does require CCR2 or its ligand.
Lipolysis induces macrophage migration. Perigonadal adipose tissue explants were isolated from lean mice that were either ad libitum fed or were fasted for 24 hours. Explants from fasted animals were incubated under basal conditions, whereas explants from ad libitum–fed mice were incubated with or without isoproterenol treatment (10 μM). (A) FFA concentration was measured in the explant-conditioned media. (B) The chemotactic activity of control medium, medium supplemented with MCP-1/CCL2 (50 ng/ml), and explant-conditioned medium were measured using a standard migration assay for BMDMs. Data are represented as mean ± SD. n = 4, 5–8 replicates per sample. *P < 0.05; **P < 0.01.
Weight loss and lipolysis activate a program of lipid uptake by ATMs. A primary function of macrophages is the phagocytosis of tissue-specific products in a manner that maintains tissue homeostasis. For example, in bone, osteoclast (the multinucleated macrophage of bone) resorption of matrix is necessary to maintain bone health (39). By analogy, the accumulation of ATMs during periods of elevated lipolysis may permit uptake or phagocytosis of excess local lipids and participate in the turnover of lipid in adipose tissue. Indeed, a distinctive characteristic of ATMs in obesity is the accumulation of intracellular lipid (19, 29). In adipose tissue from obese individuals, there are ATMs with multiple lipid droplets and others that form multinucleated giant cells that contain large unilocular droplets. Consistent with a primary function of ATMs being the uptake of lipid during periods of increased release of FFA from adipocytes, the expression in adipose tissue of 2 macrophage lipid transport receptors, Cd36 and Msr1, is increased during the initial period of weight loss (Figure 7A). By day 42 of caloric restriction, when adipose tissue lipolysis is reduced, the expression of these 2 genes returns to levels seen in the never–weight-reduced mice. To determine whether lipolysis acutely induces accumulation of lipid within ATMs, we studied lipid droplet content in macrophage-containing stromal vascular cells (SVCs) following a 24-hour fast. Fasting acutely induced lipid droplet formation in ATMs and increased the number of lipid-containing vesicles by 39% in ATMs isolated from fasting compared with ATMs from ad libitum–fed obese C57BL/6J mice (17.3 ± 3 vs. 24.1 ± 4.7 lipid vesicles per cell; P < 0.001; Figure 7, B and C). Fasting also increased the expression of genes involved in lipid uptake, storage (Adfp, aP2, Cd36), and export (Abca1 and ApoE) in the stromal vascular fraction of adipose tissue (Figure 7D).
Fasting acutely induces lipid droplet formation in ATMs. (A) Adipose tissue expression of genes whose products, CD36 (Cd36) and scavenger receptor A (Msr1), are implicated in lipid uptake by macrophages were measured in perigonadal adipose tissue during weight loss induced by caloric restriction. n = 5–6 mice/group. *P <0.05; ***P < 0.001, versus day 0. (B) Stromal vascular cells (SVCs) isolated from perigonadal adipose tissue of high-fat diet–induced obese mice that were fed ad libitum (left panel) or were fasted for 24 hours (right panel) were stained for neutral lipid with oil red O. Scale bars: 50 μm. (C) Number of lipid droplets in macrophages from perigonadal adipose tissue of high-fat diet–induced obese mice fed ad libitum or fasted for 24 hours. n = 5 mice/group. ***P < 0.001. (D) Expression of genes (in the stromal vascular fraction) encoding proteins involved in lipid uptake, utilization, and export. n = 5 mice/group. *P < 0.05. All data are represented as mean ± SD.
To more directly determine whether ATMs take up lipid in response to lipolysis per se, we established an in vitro system to assess the effects of adrenergic-stimulated adipocyte lipolysis on ATM function. We cocultured SVCs that include macrophages with adipose tissue from lean C57BL/6J mice in the presence of an adrenergic stimulus (isoproterenol). In the absence of adipose tissue, treatment with isoproterenol had no effect on SVC gene expression or lipid accumulation (Figure 8, A and C, and Supplemental Figure 10A). However, in the presence of adipose tissue, isoproterenol induced the expression of the gene Adfp that encodes the lipid droplet protein that coats lipid droplets in ATMs (19). In addition, the expression of Cd36 was increased (P = 0.09) (Figure 8A). There was no activation of a program of adipocyte differentiation, i.e., no increase in expression of adiponectin or leptin and decreased expression of Pparg (Supplemental Figure 10B). In addition, coculture with adipose tissue, and more profoundly with the addition of isoproterenol stimulation, induced Ccr2 gene expression in the SVC fraction, potentially reflecting an activation of a chemotactic macrophage program (Figure 8B). Histologically, induction of lipolysis led to the accumulation of lipid within SVCs (Figure 8, C and D) and increased the appearance of multinucleated lipid-laden macrophages typically seen in adipose tissue of obese individuals (Figure 8E).
Induction of adipose tissue lipolysis activates lipid uptake by ATMs. (A) SVCs isolated from perigonadal adipose tissue of high-fat diet–induced obese mice were cultured either alone or with perigonadal adipose tissue pieces (harvested from lean animals) with or without isoproterenol treatment (10 μM) to induce lipolysis in the adipose tissue fraction. The gene expression of Adfp and Cd36 in SVCs was measured. n = 5 mice/group. Data are represented as mean ± SD. (B) The expression of the chemokine receptor Ccr2 was measured in SVCs treated as described in A. Data are represented as mean ± SD. n = 5 mice/group. (C) SVCs treated with isoproterenol cultured alone (left panel) or with adipose tissue (right panel) were stained for neutral lipid with oil red O. Lipid-containing cells are marked with arrows. Scale bars: 50 μm. (D) Percentage of lipid-containing cells among SVCs treated as described in A. n = 5 mice/group. Data are represented as mean ± SD. *P < 0.05; **P < 0.01; ***P < 0.001; †P = 0.09. (E) The presence of lipid-laden multinucleated giant cells among SVCs cocultured with adipose tissue in the presence of isoproterenol. Scale bar: 15 μm.
Macrophage depletion increases adipose tissue lipolysis. The recruitment of ATMs and the uptake of lipid during periods of lipolysis led us to hypothesize that ATMs play a role in regulating local concentrations of FFA. To determine whether ATMs affect lipolysis or lipid release, we depleted ATMs from adipose tissue that harbored a high concentration of macrophages. Epididymal adipose tissue explants were collected from high-fat diet–induced obese mice that were fasting for 24 hours. The explants were treated with liposome-encapsulated clodronate to deplete ATMs, and control explants from the same animals were treated with liposome-encapsulated PBS. Clodronate is a biphosphonate that induces macrophage apoptosis when phagocytosed in liposomes. As expected, clodronate treatment reduced by approximately 80% the macrophage content as measured by expression of the macrophage marker Emr1 and the macrophage-expressed scavenger receptor Msr1. Macrophage depletion increased the expression of the lipase Atgl/Pnpla2 by 2.5-fold and increased the expression of fatty acid–regulated genes Fabp4/aP2, Acadl, and Dgat1. The induction of a lipase and genes required for fatty acid metabolism suggested that macrophage depletion increases lipolysis and FFA substrates. Indeed, compared with control-treated adipose tissue explants, macrophage-depleted explants had a 27% higher rate of lipolysis as measured by glycerol release. To assess whether macrophage depletion might have a similar effect on FFA metabolism in vivo, we injected lean C57BL/6J mice intraperitoneally with liposome-encapsulated clodronate to deplete ATMs from intra-abdominal adipose tissue depots and subsequently fasted them for 24 hours. Clodronate treatment compared with control (liposome-encapsulated PBS) treatment reduced macrophage-specific gene expression by more than 90% (Supplemental Figure 10C). Intra-abdominal ATM depletion increased fasting serum (FFA) by 69% compared with control liposome-injected mice (Figure 9C). These data provide preliminary evidence that macrophages function in part to attenuate lipolysis-induced release of FFA.
Adipose tissue explants were isolated from high-fat diet–induced obese mice that were fasting for 24 hours. Subsequently, explants were treated either with liposome-encapsulated clodronate or liposome-encapsulated PBS. Explants from the same mice were treated with both experimental conditions. (A) Gene expression of macrophage-specific genes and genes involved in lipid metabolism in the explants. Data are represented as mean ± SD. n = 4 mice/group. (B) Glycerol release from explants of perigonadal adipose tissue treated either with liposome-encapsulated clodronate or liposome-encapsulated PBS. Liposome-encapsulated clodronate was administered intraperitoneally to lean C57BL/6J mice. Mice were fasted for 24 hours starting on day 3 after injection, and macrophage depletion in perigonadal adipose tissue was confirmed at the end of the fasting period (day 4). Liposome-encapsulated PBS was also administered as control. (C) Serum concentration of FFA in clodronate- or PBS-treated mice after a 24-hour fast. Data are represented as mean ± SD. n = 8 mice/group. *P < 0.05; **P < 0.01, versus PBS treated.