Impact of Reduced ATGL-Mediated Adipocyte Lipolysis on Obesity-Associated Insulin Resistance and Inflammation in Male Mice

Emerging evidence suggests that impaired regulation of adipocyte lipolysis contributes to the proinflammatory immune cell infiltration of metabolic tissues in obesity, a process that is proposed to contribute to the development and exacerbation of insulin resistance. To test this hypothesis in vivo, we generated mice with adipocyte-specific deletion of adipose triglyceride lipase (ATGL), the rate-limiting enzyme catalyzing triacylglycerol hydrolysis. In contrast to previous models, adiponectin-driven Cre expression was used for targeted ATGL deletion. The resulting adipocyte-specific ATGL knockout (AAKO) mice were then characterized for metabolic and immune phenotypes. Lean and diet-induced obese AAKO mice had reduced adipocyte lipolysis, serum lipids, systemic lipid oxidation, and expression of peroxisome proliferator-activated receptor alpha target genes in adipose tissue (AT) and liver. These changes did not increase overall body weight or fat mass in AAKO mice by 24 weeks of age, in part due to reduced expression of genes involved in lipid uptake, synthesis, and adipogenesis. Systemic glucose and insulin tolerance were improved in AAKO mice, primarily due to enhanced hepatic insulin signaling, which was accompanied by marked reduction in diet-induced hepatic steatosis as well as hepatic immune cell infiltration and activation. In contrast, although adipocyte ATGL deletion reduced AT immune cell infiltration in response to an acute lipolytic stimulus, it was not sufficient to ameliorate, and may even exacerbate, chronic inflammatory changes that occur in AT in response to diet-induced obesity.

Obesity is a global public health problem that is highly associated with insulin resistance, diabetes, fatty liver disease, and cardiovascular disease. An early characteristic of these disorders is accumulation of lipids within multiple tissues, usually in association with adipocyte dysfunction, enhanced adipocyte lipolysis, and increased serum lipids. Specific lipid species have been shown to promote cellular toxicity (lipotoxicity) via a variety of mechanisms (1). Of particular interest is the role of intra- and extracellular lipids in promoting the inflammatory response and, more specifically, the recruitment and activation of immune cells into metabolically relevant tissues, including liver and adipose tissue (AT). Immune cell recruitment and activation occur not only in obesity, where insulin-mediated inhibition of lipolysis is impaired (2) and release of lipids from dead/dying adipocyte is enhanced (3, 4) but also in other prolipolytic states, including weight loss (5), fasting (5), and β₃-adrenergic stimulation (5, 6). This raises the important question of whether modulation of adipocyte lipolysis might promote or protect against obesity-associated inflammatory responses. Understanding the mechanisms by which lipid excess and/or production contribute to these inflammatory responses may lead to novel strategies to prevent or treat metabolic disease.

The rate-limiting enzyme regulating mobilization of fatty acids (FAs) from intracellular triacylglycerol (TAG) stores is adipose triglyceride lipase (ATGL) (7–9). ATGL is expressed in essentially all tissues, particularly adipocytes, where it promotes both basal and stimulated lipolysis. Not surprisingly, global ATGL deletion dramatically reduces TAG hydrolysis in adipose and non-ATs, resulting in impaired release of FAs both locally and systemically (10). This reduced lipolysis improves glucose homeostasis and produces tissue-specific changes in insulin action (10, 11). Whether these metabolic changes are accompanied by changes in immune phenotypes in metabolically relevant tissues remains unknown. On the one hand, ATGL action has been implicated in the recruitment of immune cells to AT in response to acute lipolytic stimuli (5). In support of this hypothesis, global ATGL knockout (GAKO) mice have reduced AT macrophage (ATM) infiltration after prolonged fasting (5). On the other hand, ATGL action has been proposed to protect against the AT immune response to nutritional and age-related stress by producing FAs that not only provide energy but also serve as ligands for peroxisome proliferator-activated receptors (PPARs) (12, 13), key nuclear transcription factors known to influence both metabolism and inflammation (14). In support of this hypothesis, GAKO mice have increased mRNA expression of inflammatory cytokines (ie, Tnfa, Il6) in AT (12, 13). To further complicate the matter, ATGL is also expressed in macrophages where it is required for normal macrophage function, including migration, phagocytosis, and survival (15–17). However, studies in adipocyte-specific ATGL knockout (AAKO) mice (18, 19) have not evaluated immune phenotypes and additionally may be confounded by use of the adipocyte fatty acid binding protein 4 (aP2)-promoter for Cre-mediated gene deletion, a promoter that may drive Cre expression in tissues other than adipocytes, including macrophage (20, 21). Thus, the relative contributions of adipocyte vs macrophage lipolysis to immune cell recruitment as well as the role of adipocyte-specific ATGL action in metabolic and immune responses to obesity remain unclear.

The goal of the present study was to determine the impact of reduced ATGL-mediated adipocyte lipolysis on metabolic and immune responses to diet-induced obesity (DIO). To achieve this goal, we generated mice with adipocyte-specific targeted deletion of ATGL using the adiponectin (Adipoq) rather than aP2 promoter for Cre-mediated ATGL deletion, because it drives Cre expression more specifically in mature adipocytes and not myeloid or other cells (20, 21). Because this is the first report of congenital AAKO mice generated using Adipoq-Cre-mediated ATGL deletion, a secondary goal of the present study was to report the basic metabolic phenotype of this model. We found that inhibition of ATGL-mediated adipocyte lipolysis protects against hepatic steatosis and immune cell infiltration/activation in association with improved hepatic insulin sensitivity and systemic glucose homeostasis. In contrast, although inhibition of ATGL-mediated adipocyte lipolysis reduces immune cell recruitment into AT in response to acute lipolysis, it was not sufficient to ameliorate, and may even exacerbate, the chronic inflammatory changes that occur in AT in response to DIO.

Materials and Methods

Animals

B6N.129-Pnpla2^tm1Eek (ATGL-flox) mice (22) were bred to B6N.FVB.Tg(Adipoq-Cre)^1evr/J mice (23) to generate AAKO mice. Mice were backcrossed onto C57BL/6NTac for >N10. Male ATGL^flox/flox Cre/+ mice were mated to female ATGL^flox/flox +/+ mice to generate ATGL^flox/flox Cre/+ (AAKO) and ATGL^flox/flox +/+ (control) mice. Genotyping was performed using established protocols (22). Mice were housed under standard conditions (25°C, 14-h light, 10-h dark cycle) with ad libitum access to chow (14 kcal% fat, Isopro RMH3000; Prolab) or high-fat diet (HFD) (45 kcal% fat, D12451; Research Diets). Diets were initiated after weaning at 3 weeks of age. For experiments performed under fasting conditions, mice were fasted in the dark phase from 9 pm to 9 am. Body composition, energy expenditure, and metabolic parameters were determined as described (11, 22, 24). Experiments were approved by the University of Pittsburgh Institutional Animal Care and Use Committee and performed in conformity with Public Health Service Policy for Care and Use of Laboratory Animals.

Tissue imaging

Antibodies for imaging are summarized in Supplemental Table 1. For immunofluorescence (IF) of ATGL in AT, 5-μm sections of formalin-fixed brown AT (BAT) were stained for ATGL (1:100, 2439; Cell Signaling) followed by donkey antirabbit IgG secondary antibody (1:200, A21206; Invitrogen) (22). For IF of ATGL and F4/80 in cultured peritoneal macrophage, mice were injected ip with 2 mL of 6-g/L thioglycolate followed by collection of peritoneal cells 72 hours thereafter. Resulting cells were cultured for 24 hours in complete DMEM containing 10% lipoprotein-deficient serum (BT-907; Biomedical Technologies) to enrich for macrophage. Cells were fixed in 4% paraformaldehyde (PFA) and stained for ATGL as noted above as well as F4/80 (1:100, ab6640; Abcam) followed by rabbit antirat IgG secondary antibody (1:200, T4280; Sigma-Aldrich). For AT morphology, 5-μm sections of formalin-fixed AT were stained with hematoxylin and eosin (H&E). Adipocyte size distribution was determined by measuring area of 300 adipocytes in 3 random areas. For liver neutral lipid staining, 10-μm sections of isopentane-frozen liver were stained with Oil red O (22). For IF of CD11c in liver, 5-μm sections of formalin-fixed liver were stained for CD11c (1:100, 550283; BD Pharmingen) followed by goat antirat IgG secondary antibody (1:500, A011006; Invitrogen). For immunohistochemistry of AT, 5-μm sections of formalin-fixed AT were stained for CD11c (1:100, CL8942AP; Cedarlane) or F4/80 (1:100, ab6640; Abcam) followed by goat antirat IgG secondary antibody (1:500, 112-035-003; Jackson ImmunoResearch) and 3,3′-diaminobenzidine substrate (SK-4100; Vectorlab) (4, 17, 25). Images were visualized by light microscopy (DM4000B; Leica), digitally captured (Retiga 2000R camera; QImaging), and analyzed using Northern Eclipse 6.0 software (Empix Imaging). For IF of CD11c and F4/80 in AT, whole-mount PFA-fixed AT was stained as described (25, 26) using the above antibodies with the exception that goat antihamster IgG secondary (1:1000, 127-165-160; Jackson ImmunoResearch) was used for F4/80. Nuclei were stained with Hoechst stain (33258; Sigma). Confocal stack reconstructions (50 μm at 5-μm intervals) were generated using an Olympus Fluoview 1000 Microscope (Olympus) and Imaris software (Bitplane).

Flow cytometry and cytokine determination

Antibodies for flow cytometry are summarized in Supplemental Table 1. Stromal vascular cells (SVCs) of white AT (WAT) (perigonadal plus retroperitoneal) and mononuclear cells (MCs) from liver were isolated using established protocols (25). For flow cytometry, 2 × 10⁶ cells per sample/tissue were incubated with antimouse CD16/CD32 (14-0161-86; eBioscience) for 10 minutes at 4°C. Cell surface staining was performed in 1× PBS containing 2% fetal bovine serum with anti-CD45-PerCP (557235; BD Biosciences), anti-CD11b-PeCy7 (552850; eBioscience), anti-CD11c-APC (550261; eBiosciences), and anti-F4/80-Alexa Fluor 780 (474801; eBiosciences) for 30 minutes at 4°C. Cells were washed and resuspended in 4% PFA and analyzed using FACSCalibur flow cytometer and FACSDiva software (BD Biosciences). To measure cytokine and chemokine secretion, MCs from liver were incubated with complete DMEM containing 10% heat-inactivated fetal bovine serum, nonessential amino acids, and 2-mercaptoethanol for 24 hours, and the medium was assayed by Luminex (TNFα; Millipore) or ELISA (MCP1 and IL-6; R&D Systems).

Protein and gene expression

Protein expression and/or phosphorylation under basal (saline) and insulin-stimulated (0.7 U/kg) condition was performed as described (11) using the next primary antibodies (see also Supplemental Table 1): anti-Akt^pThr308 (1:250, 9275S; Cell Signaling), anti-Akt^pSer473 and anti-Akt^total (1:250, 05-736 and 07-416; EMD Millipore) and anti-Ran GTPase (1:10 000, 610340; BD Biosciences). Immunoblots were developed using Immun-StarWestern C kit (Bio-Rad) and quantified with a GelDoc System and Quantity One 1-D software (Bio-Rad). For gene expression analysis, RNA was isolated from frozen tissue or cultured peritoneal macrophage (RNeasy Lipid Tissue Mini Kit; QIAGEN). cDNA was synthesized from RNA (qScript Supermix; Quanta Biosciences). mRNA expression was determined by qPCR (PerfeCTa fastmix; Quanta Biosciences) using an Eppendorf Realplex System using primers listed in Supplemental Table 2 in accordance with guidelines for minimal information for publication of quantitative real-time PCR experiments (11, 22, 24) (for antibodies, see Table 1).

Table 1.

Antibody Table

Antibody	Peptide/Protein Target	Antigen Sequence (if known)	Name of Antibody	Manufacturer, Catalog Number, and/or Name of Individual Providing the Antibody	Species Raised in; Monoclonal or Polyclonal	Dilution Used
Primary	ATGL		ATGL (30A4) rabbit mAb	Cell Signaling, 2439	Rabbit; monoclonal	1:100
Secondary	N/A		Alexa Fluor 488 donkey antirabbit IgG (H + L) antibody	Life Technologies, A-21206	Donkey; polyclonal	1:200
Primary	F4/80		Anti-F4/80 antibody [CI: A3-1]	Abcam, ab6640	Rat; monoclonal	1:100
Secondary	N/A		Antirat IgG (whole molecule)-TRITC antibody	Sigma-Aldrich, T4280	Rabbit; polyclonal	1:200
Primary	CD11c		Hamster antimouse CD11c	BD Pharmingen, 550283	Hamster; monoclonal	1:100
Secondary	N/A		Alexa Fluor 488 goat antirat IgG (H + L) antibody	Life Technologies, A-11006	Goat; polyclonal	1:500
Primary	CD11c		Antimouse/human Mac-2 (Galectin-3), purified (clone M3/38) (rat IgG2a)	Cedarlane Laboratory, CL8942AP	Rat; monoclonal	1:100
Secondary	N/A		Antirat IgG (H + L)	Jackson ImmunoResearch, 112-035-003	Goat; polyclonal	1:500
Secondary	N/A		Cy3-AffiniPure goat anti-Armenian hamster IgG (H + L)	Jackson ImmunoResearch, 127-165-160	Goat; polyclonal	1:1000
Primary	Akttotal	NP_001014421	Anti-Akt1/PkBα antibody	EMD Millipore, 07-416	Rabbit; polyclonal	1:250
Primary	AktpSer473		Antiphospho-Akt1/PKBα (Ser473) antibody	EMD Millipore, 05-736	Rabbit; monoclonal	1:250
Primary	AktpThr308		Antiphospho-Akt1/PKBα (Thr308) antibody	Cell Signaling, 9275S	Rabbit; polyclonal	1:250
Primary	RAN GTPase	Human RAN, aa 7-171	Ran	BD Biosciences, 610340	Rat; monoclonal (clone 20/Ran)	1:10 000
Primary	CD16/CD32		Antimouse CD16/CD32	eBioscience, 14-0161-86	Rat; monoclonal (clone 93)	N/A
Primary	CD45		Anti-CD45-PerCP	BD Biosciences, 557235	Rat; monoclonal (clone 30-F11)	N/A
Primary	CD11b		Anti-CD11b-PeCy7	eBioscience, 552850	Rat; monoclonal (clone M1/70)	N/A
Primary	CD11c		Anti-CD11c-APC	eBioscience, 550261	Rat; monoclonal (clone HL3)	N/A
Primary	F4/80		Anti-F4/80-Alexa Fluor 780	eBioscience, 474801	Rat; monoclonal (clone BM8)	N/A

Statistical analysis

Results are expressed as mean ± SEM. Comparisons were made by unpaired 2-tailed Student's t test or factorial ANOVA followed by determination of simple effects for pair-wise comparisons if relevant. For repeated measurements, comparisons were made by two-way ANOVA with repeated measures. For comprehensive laboratory animal monitoring system data, comparisons were made using generalized estimate equations. For all analyses, P < .05 were considered statistically significant.

Results

Adipocyte ATGL deletion reduces adipocyte lipolysis, serum lipids, and lipid oxidation

AAKO mice were generated by crossing ATGL-flox mice (22) with mice expressing Cre recombinase under the control of the Adipoq promoter (23). AAKO mice had dramatically reduced Atgl mRNA (Figure 1A) and protein (Figure 1B) expression exclusively in AT but not in non-ATs or cultured peritoneal macrophages. Adipocyte ATGL deletion dramatically reduced basal and stimulated lipolysis, both in vivo (Figure 1C) and in AT explants ex vivo (Figure 1D). As a result, both chow- and HFD-fed AAKO mice had lower serum non-esterified fatty acids (NEFAs), TAGs, and ketone bodies (Table 2). Adipocyte ATGL deletion also reduced systemic oxidation of lipid relative to carbohydrate substrates, as indicated by increased respiratory exchange ratios in both ad libitum-fed and fasted states (Figure 1E). In addition, mRNA expression of PPARα, PPARα target genes, and genes involved in lipid oxidation was lower in perigonadal WAT, BAT, and liver of ad libitum-fed 24-week-old AAKO mice fed chow and HFD (Figure 1F). Thus, adipocyte-specific ATGL deletion reduces adipocyte lipolysis, serum lipids, systemic lipid oxidation, and expression of PPARα and PPARα target genes involved in lipid oxidation in AT and liver.

Effects of adipocyte-specific ATGL deletion on adipocyte lipolysis, serum lipids, lipid oxidation, and oxidative gene expression

Figure 1.

A, left panel, Atgl mRNA expression relative to cyclophilin reference gene by qPCR in AT (BAT, brown; PGAT, perigonadal; SCAT, sc; MAT, mesenteric) and non-ATs (Gas, gastrocnemius) with endogenous Atgl expression in control BAT arbitrarily set to 1 (M, 8–9 wk, chow, fasted 12 h, n = 4–5/group). Right panel, Atgl mRNA expression in cultured peritoneal macrophages with expression in control macrophage arbitrarily set to 1 (M, 12 wk, fasted 12 h, n = 5–6/group). B, left panel, IF of ATGL (green) in BAT. Right panels, IF of ATGL alone (green) and in combination with the macrophage marker F4/80 (red; merge of ATGL+F4/80 is yellow/orange) in cultured peritoneal macrophages (M, 8–12 wk, chow, fasted 12 h, representative images). C, In vivo lipolysis as determined by serum NEFAs at time 0 (basal) and 15 minutes (stim, stimulated) after ip administration of 1-mg/kg body weight (BW) of the β₃-adrenergic receptor agonist CL 316,243 (M, 10 wk, chow, fasted 6 h, n = 9–10/group). D, Ex vivo lipolysis determined by NEFA released from PGAT explants incubated without (basal) or with (stim) 10μM β-adrenergic receptor agonist isoproterenol (F, 12 wk, fasted 12 h, n = 6–7/group). E, Respiratory exchange ratio (RER) in weight-matched chow- and HFD-fed mice (M, 8–9 wk, n = 4/group). F, mRNA expression relative to cyclophilin reference gene by qPCR with the control-chow group arbitrarily set to 1 for each gene (M, 24 wk, ad libitum fed, n = 4–8/group); P < .05: &, effect of treatment; *, effects of genotype; and #, effects of diet.

Table 2.

Blood and Tissue Metabolites in Control vs AAKO Mice Fed Chow or HFD

Parameter	Control Chow	AAKO Chow	Control HFD	AAKO HFD
NEFA (mEq/L)	0.55 ± 0.06	0.21 ± 0.04a	0.48 ± 0.03	0.18 ± 0.02a
TAG (mg/mL)	1.01 ± 0.12	0.17 ± 0.02a	1.13 ± 0.10	0.17 ± 0.02a
Ketone bodies (mM)	1.55 ± 0.19	0.49 ± 0.17a	1.52 ± 0.11	0.39 ± 0.03a
Cholesterol (mg/mL)	0.83 ± 0.06	0.62 ± 0.07	1.20 ± 0.09b	0.93 ± 0.08a,b
Glucose (mg/dL)	83 ± 2.8	87 ± 4.2	111 ± 5.2b	91 ± 3.6a
Insulin (ng/mL)	0.58 ± 0.10	0.10 ± 0.06a	1.82 ± 0.28b	0.10 ± 0.06a

Blood energy metabolites in control vs AAKO mice fed chow and HFD (M, 24 wk, fasted 12 h, n = 5–19/group).

^a

P < .05 for effect of genotype (ie, AAKO vs control within the same diet group).

^b

P < 0.05 for effect of diet (ie, chow vs HFD within the same genotype group).

Adipocyte ATGL deletion alters regional fat distribution without expansion of overall fat mass, in part, by reducing expression of genes involved in adipocyte lipid uptake, lipid synthesis, and adipogenesis

Given the decrease in lipolysis and lipid oxidation, we expected AAKO mice to have increased fat mass. Instead, adipocyte-specific ATGL deletion produced only a small, transient increase in body weight (Figure 2A) and total fat mass (Figure 2B) in mice fed chow and HFD over 24 weeks, similar to ATGL-deficient humans (27, 28). Consistent with these results, differences in total energy intake and expenditure (estimated by oxygen consumption [VO₂]) at 8–9 weeks (Figure 2C) and 20–21 weeks (data not shown) could not be detected between genotypes in the ad libitum-fed state, even though VO₂ was drastically reduced in AAKO mice in the fasted state. Although AAKO mice did not differ from controls in total fat mass (Figure 2B) at 24 weeks of age, they did exhibit differences in regional fat distribution characterized by an increase in BAT mass and depot-specific decreases in WAT mass (ie, lower sc fat in HFD-fed mice) (Figure 2D). Gross and microscopic morphology of BAT from chow-fed AAKO mice revealed a “WAT-like” appearance with large unilocular lipid droplets in brown adipocytes (Figure 2E, left), whereas perigonadal WAT morphology by H&E appeared relatively similar to chow-fed control mice (Figure 2E, right). Despite similar appearance, perigonadal adipocyte size tended to be smaller in chow- but not HFD-fed mice (Figure 2F). Furthermore, liver weight and fat (TAG) content were markedly reduced in AAKO mice, as was liver glycogen content (Figure 2, G and H). The reduction in hepatic fat was significant in both fasted (decreased 4.9- and 6.9-fold in chow and HFD-fed AAKO mice, respectively) and ad libitum-fed mice (decreased 2.5- and 5.7-fold in chow and HFD-fed AAKO mice, respectively). Consistent with these results, hepatic mRNA expression of genes involved in TAG synthesis (ie, GK and Dgat2) and de novo lipogenesis (Fasn) were either decreased or unchanged, respectively, whereas genes involved in hepatic gluconeogenesis were markedly increased (Figure 2I). Thus, adipocyte-specific ATGL deletion alters fat distribution without an overall expansion of fat mass by 24 weeks of age.

Effects of adipocyte-specific ATGL deletion on systemic energy homeostasis, fat distribution, and fat synthesis/storage

Figure 2.

Longitudinal body weight (A) and fat mass (B) by EchoMRI in control and AAKO mice fed chow and HFD (M, 3–24 wk, ad libitum fed, n > 20/group). C, Energy intake (left; M, 8–9 wk, ad libitum fed, n > 20/group) and expenditure (right; M, 8–9 wk, n = 4/group) in chow- and HFD-fed control and AAKO mice. D, Fat pad weights (M, 24 wk, fasted 12 h, n > 20/group). E, Gross and microscopic morphology by H&E of BAT (left) and perigonadal WAT (pgWAT, right) of control and AAKO mice (M, 24 wk, chow and HFD, representative images). F, Adipocyte size distribution of pgWAT fed chow (top) or HFD (bottom) (M, 24 wk). G, Liver weight, TAG content, and glycogen content (M, 24 wk, fasted 12 h, n > 20/group). H, Liver gross images and neutral lipid content by Oil red O (ORO) (M, 24 wk, chow and HFD, representative images). I–K, Liver (I), WAT (J), and BAT (K) mRNA expression relative to cyclophilin reference gene by qPCR with the control-chow group arbitrarily set to 1 for each gene (M, 24 wk, ad libitum fed, n = 4–8/group). P < .05: *, effects of genotype and #, effects of diet.

Our recent collaborative work has demonstrated that chronic β₃-adrenergic-stimulated adipocyte lipolysis is not only coupled to lipid oxidation but also to lipid synthesis, an effect that is abolished by inducible ATGL ablation in adipocytes (29). We therefore hypothesized that the lack of progressive AT mass expansion in AAKO mice might result in part from decreased adipocyte lipid synthesis and/or similar mechanisms that prevent lipid storage in adipocytes. Indeed, in ad libitum-fed 24-week-old AAKO mice, mRNA expression of PPARγ2 (but not PPARγ1) and other key transcriptional regulators of lipogenesis and/or adipogenesis (ie, Cebpa, Srebp1c) was reduced in WAT of HFD-fed AAKO mice (Figure 2J) as well as in BAT of both chow- and HFD-fed AAKO mice (Figure 2K). In addition, mRNA expression of PPARγ target genes and other genes involved in lipid uptake (Fabp4, Cd36) and synthesis (Gyk, Pepck, Fasn, Agpat2, Dgat2) was decreased in WAT (Figure 2J) and BAT (Figure 2K) of AAKO mice on both diets. Notably, these effects were present in BAT despite an overall increased BAT mass, suggesting that expansion resulting from impaired lipolysis and/or lipid oxidation may subsequently be restrained by mechanisms limiting further lipid storage. Thus, these data suggest that impaired ATGL action in adipocytes not only decreases PPARα-mediated lipid oxidation but also PPARγ-mediated processes such as lipid uptake, lipid synthesis, and adipogenesis.

Adipocyte ATGL deletion improves systemic glucose homeostasis, at least in part, by improving hepatic insulin signaling

Because changes in adipocyte biology, systemic lipid homeostasis, and hepatic fat content are major contributors to insulin action, we next assessed the effects of adipocyte-specific ATGL deletion on systemic glucose homeostasis and tissue-specific insulin signaling. AAKO mice on both diets had reduced blood glucose after a 6-hour fast (time 0 of glucose tolerance test; Figure 3A) and were protected from diet-induced increases in fasting blood glucose and serum insulin (Table 2). Serum insulin concentrations were also very low in AAKO mice independent of diet (Table 2). AAKO mice on both diets had enhanced glucose clearance in response to both glucose (Figure 3A) and insulin challenges (Figure 3B). These data suggest that AAKO mice have improved systemic glucose homeostasis and insulin action. Examination of the tissue-specific contributions of insulin signaling to these systemic phenotypes revealed a dramatic increase in insulin-stimulated phosphorylation of Akt^pSer473 and Akt^pThr308 in liver of AAKO mice on both diets (Figure 3C). In contrast, no genotype effects on phosphorylation of these key insulin signaling proteins were identified in WAT (Figure 3D), BAT (data not shown), or skeletal muscle (Figure 3E). Thus, improved glucose homeostasis in AAKO mice is mediated, at least in part, by enhanced insulin signaling in liver but not AT.

Effects of adipocyte-specific ATGL deletion on systemic glucose homeostasis and tissue-specific insulin signaling

Figure 3.

A, Glucose tolerance test (GTT) of chow-fed (left) and HFD-fed (right) mice with glucose 1.6-g/kg glucose (M, 16 wk, fasted 6 h, n=11/group). B, Insulin tolerance test (ITT) of chow-fed (left) and HFD-fed (right) mice with insulin 0.75 U/kg (M, 20 wk, fasted 4 h, n = 11/group). Insets for A and B, AUC (×10³) for glucose. C–E, Insulin signaling in liver (C), perigonadal WAT (D), and skeletal muscle (tibialis anterior) (E) of AAKO mice. Mice were injected ip with saline or insulin at 0.7 U/kg of body weight (BW) and killed 10 minutes thereafter (M, 24 wk, fasted 12 h, n = 4–7/group). Quantification of stoichiometric phosphorylation of Akt pSer⁴⁷³/Akt total and Akt pThr³⁰⁸/Akt total (left panel) and total Akt/Ran GTPase control (middle panel) as well as representative immunoblots (right panel). P < .05: *, effects of genotype. Overall effects of diet are not shown but are as follows: liver Akt pThr³⁰⁸/Akt, P = .053; PGAT Akt pSer⁴⁷³/Akt, P < .05 and Akt pThr³⁰⁸/Akt, P < .05; skeletal muscle Akt pSer⁴⁷³/Akt, P = .08.

Adipocyte ATGL deletion attenuates infiltration and activation of immune cells in liver in response to DIO

Because lipid-induced changes in hepatic inflammation influence glucose homeostasis and insulin signaling (30–33), we next investigated how chronic reduction of adipocyte lipolysis might influence immune cell infiltration and activation in liver. CD11c is the main marker for cells of the monocyte lineage in liver. Furthermore, we have previously established that CD11c+ cells are highly recruited to the liver in response to HFD feeding, whereas CD11b+CD11c+F4/80+ (triple+) cells are unaffected (25). Consistent with our previous work, DIO increased hepatic mRNA expression of Cd11c in liver (Figure 4A). However, this HFD-induced increase in hepatic Cd11c was completely prevented in AAKO mice (Figure 4A). IF analysis confirmed accumulation of CD11c+ immune cells in liver of HFD-fed control but not AAKO mice (Figure 4B). These results were further corroborated using flow cytometry to demonstrate the expected HFD-induced increase in several subpopulations of hepatic macrophage and dendritic cells (CD11b+ and/or CD11c+ cells) in HFD-fed control but not AAKO mice (Figure 4C and Supplemental Figure 1). These changes in immune cell markers were also associated with changes in inflammatory cytokines. Specifically, the HFD-induced increase in mRNA expression of several proinflammatory cytokines (Mcp1, Tnfa, IL-6) in control mice was absent in AAKO mice, whereas mRNA expression of several antiinflammatory genes (IL-10, Arg1) was higher in AAKO mice (Figure 4D). Consistent with these findings, immune cell secretion of the proinflammatory proteins MCP1 and IL-6 (but not TNFα) was increased in HFD-fed control but not AAKO mice (Figure 4E). Thus, adipocyte ATGL deletion reduces hepatic lipid content and attenuates diet-induced immune cell infiltration and inflammation in liver.

Effects of adipocyte-specific ATGL deletion on liver immune cell infiltration and inflammation

Figure 4.

A, Hepatic mRNA expression of immune cell markers relative to cyclophilin reference gene by qPCR with the control-chow group arbitrarily set to 1 for each gene (M, 24 wk, ad libitum fed, n = 4–8/group). B, IF imaging of CD11c in liver (M, 24 wk, representative images). C, Flow cytometry of MCs isolated from liver demonstrating the total number of MCs and the fold change in CD11b+, CD11c+, CD11b+CD11c+, CD11b−CD11c+, and CD11b+CD11c− cells normalized to control-chow (M, 24–26 wk, ad libitum fed, n = 6–7/group). F4/80+ cells are not shown due to relatively low numbers in liver. Representative flow cytometry plots from liver are also shown. D, Hepatic mRNA expression of cytokines relative to cyclophilin reference gene by qPCR with the control-chow group arbitrarily set to 1 for each gene (M, 24 wk, ad libitum fed, n = 4–8/group). E, MCP1, IL-6, and TNFα protein secretion from MCs isolated from liver (M, 24–26 wk, ad libitum fed, n = 4/group). P ≤ .05: *, effects of genotype and #, effects of diet.

Adipocyte ATGL deletion attenuates infiltration of immune cells in AT in response to acute lipolytic stimuli

Given the above data and the well-established role of AT inflammation in insulin resistance and metabolic dysregulation, we next sought to understand how modulation of adipocyte lipolysis might affect AT immune phenotypes. In AT, acute activation of lipolysis has been shown to enhance recruitment of ATM to WAT (5, 6), a process that is reduced in GAKO mice (5). To further test this hypothesis in AAKO mice, we investigated the contribution of adipocyte-specific ATGL action to immune cell infiltration into AT in response to acute lipolytic stimuli. As expected, acute stimulation of lipolysis by the β₃-adrenergic agonist CL 316,243 and a 24-hour fast dramatically increased formation of crown-like structures in perigonadal WAT of control but not AAKO mice (Figure 5, A and B). These results were corroborated by a corresponding increase in mRNA expression of Cd11c (Figure 5C) and F4/80 (Figure 5D) in control but not AAKO mice. These data support a role for ATGL-mediated TAG hydrolysis in the recruitment of immune cells to AT in response to acute lipolytic challenges and also indicate that inhibition of adipocyte-specific ATGL-mediated lipolysis can prevent this response. Despite this effect, however, crown-like structures and expression of both CD11c and F4/80 were unexpectedly increased in AAKO compared with control mice in the baseline fed state.

Immune cells infiltration into AT in response to acute lipolytic stimuli

Figure 5.

Ad libitum-fed mice were injected ip with either saline (fed) or 1-mg/kg CL 316,243 (CL) at times 0 and 4 hours followed by tissue collection at 14 hours. Concurrently, a separate group of saline-treated mice were fasted for 24 hours (fast). A, Immunohistochemical staining with CD11c to assess for crown-like structures (CLSs) in perigonadal WAT (M, 40 wk, representative images). Similar results were obtained with F4/80 staining. B, Quantification of CLS (M, 40 wk, 3–4/group). The number of CLSs per high power field (hpf) were counted in 4 random sections per mouse). C and D, AT mRNA expression of Cd11c (C) and F4/80 (D) relative to cyclophilin reference gene by qPCR with expression of each gene in control-chow arbitrarily set to 1 (M, 40 wk, n = 3–5/group). P < .05: &, effect of treatment (CL or fast) and *, effect of genotype.

Chronically reducing adipocyte lipolysis via ATGL deletion does not prevent, and may even exacerbate, AT immune cell infiltration and inflammation

Based on the above results, we next assessed the AT immune response to DIO. As expected, DIO increased mRNA expression Cd11c, F4/80, and Cd68, key markers for ATMs and dendritic cells, in both BAT (Figure 6A, left) and perigonadal WAT (Figure 6B, left). However, again, instead of being decreased in AAKO mice, these markers were higher (or tended to be higher) in both BAT and WAT of AAKO mice (P < .05 for overall effect of genotype for Cd11c, F4/80, and Cd68, in WAT and BAT of AAKO mice on both diets). Likewise, AAKO mice were not protected from HFD-induced formation of CD11c+ and F4/80+ cells into crown-like structures which were equally apparent in both genotypes (Figure 6C, bottom). Furthermore, WAT of chow-fed AAKO mice demonstrated increased infiltration of CD11c+ cells in a diffuse pattern (Figure 6C, top) in addition to the more characteristic crown-like structures (Figure 5A). In a separate cohort of mice, these results were further corroborated using flow cytometry of WAT (combining perigonadal + retroperitoneal WAT), which again demonstrated increases in several subpopulations of CD11c+ cells in AT, particularly of chow-fed AAKO mice (Figure 6D and Supplemental Figure 2). Furthermore, these alterations in AT immune cells were accompanied by higher mRNA expression of several proinflammatory genes in both BAT (Figure 6A, right) and WAT (Figure 6C, right) of chow- and HFD-fed AAKO mice. Expression of several antiinflammatory genes (IL-10 and Arg1) was also higher in BAT (Figure 6, A and B). Notably, increases in mRNA expression of inflammatory cytokine were evident even in WAT of HFD-fed AAKO mice in which corresponding increases in immune cell markers were relatively less pronounced. Thus, inhibition of ATGL-mediated lipolysis in adipocytes does not prevent, and may even exacerbate, immune cell infiltration and inflammation in AT.

Effects of adipocyte-specific ATGL deletion on AT immune cell infiltration and inflammation

Figure 6.

A and B, mRNA expression of inflammatory markers (left) and cytokines (right) relative to cyclophilin control gene by qPCR in BAT (A) and perigonadal WAT (B) with the control-chow group arbitrarily set to 1 for each gene (M, 24 wk, ad libitum fed, n = 4–8/group). C, Perigonadal WAT IF for CD11c (red), F4/80 (green), and nuclei (blue) (representative images). White arrows show clustering of immune cells in crown-like structures. D, Flow cytometry of SVCs isolated from AT (combined perigonadal and retroperitoneal) demonstrating the total number of SVCs per gram of AT and the fold change in total cell number of CD11b+, CD11c+, CD11b+CD11c+, CD11b+CD11c+F4/80+ (triple+), CD11b−CD11c+, and CD11b+CD11c− cells normalized to control-chow (M, 24–26 wk, ad libitum fed, n = 6–7/group). Representative flow cytometry plots of CD11b+CD11c+ cells and CD11b+CD11c+F4/80+ cells from AT are also shown. P ≤ .05: *, effects of genotype and #, effects of diet.

Discussion

In this study, we characterized the impact of ATGL-mediated adipocyte lipolysis on the metabolic and immune response to DIO using an Adipoq-Cre adipocyte-specific ATGL knockout mouse model. We found that adipocyte-specific targeted deletion of ATGL 1) reduces adipocyte lipolysis, serum lipids, lipid oxidation, and expression of PPARα and PPARα target genes; 2) does not cause progressive fat mass expansion (by 24 wk of age) in part due to reduced PPARγ and PPARγ-mediated processes of lipid uptake, lipid synthesis, and adipogenesis; 3) improves systemic glucose homeostasis and tissue-specific insulin signaling in liver but not AT; 4) attenuates hepatic recruitment and activation of immune cells into liver in response to DIO; 5) attenuates infiltration of immune cells into AT in response to an acute lipolytic challenge; but 6) does not prevent and may even exacerbate immune cell infiltration and activation in both WAT and BAT in response to DIO. Taken together, these data suggest that inhibition of ATGL-mediated adipocyte lipolysis protects against diet-induced metabolic dysfunction in part by reducing hepatic steatosis and immune cell recruitment/activation, thereby improving hepatic insulin signaling. These data also suggest that although inhibition of ATGL-mediated adipocyte lipolysis reduces the AT inflammatory response to acute lipolysis, it is not sufficient to counteract the chronic inflammatory changes in AT that occur in the context of DIO. Instead, chronic inhibition of ATGL-mediated adipocyte lipolysis likely leads to fundamental changes in adipocyte biology that ultimately promote adipocyte dysfunction and inflammation. Thus, modulation of ATGL action produces complex tissue-specific changes in both metabolic and immune phenotypes.

A unique aspect of our study is that we generated an animal model with reduced adipocyte lipolysis by deleting ATGL exclusively from mature adipocytes (Adipoq-AAKO mice) using the Adipoq promoter to drive Cre expression (23). Two previous groups have reported metabolic effects of adipocyte ATGL deficiency (aP2-AAKO mice) (18, 19) using the aP2 promoter to drive Cre expression (34). However, these studies focused on BAT phenotypes (18) and fasting energy homeostasis (19) rather than immune phenotypes. Although aP2-Cre mice have traditionally been used to generate adipocyte-specific knockout mice, the aP2 promoter may also drive gene expression in nonadipocyte cells both during embryonic development and in adulthood (20, 21), thereby potentially confounding results. In contrast, the Adipoq promoter is more specific for mature adipocytes and has greater and more uniform expression in WAT (20, 21). Furthermore, unlike other studies (18, 19), we specifically confirmed preservation of ATGL mRNA expression as well as ATGL protein coexpression in cultured peritoneal macrophages. This confirmation is particularly important because several studies support a role for ATGL action in macrophage function that could also confound results (15–17). Thus, Adipoq-AAKO mice represent an ideal model to assess the metabolic and immune effects of altered adipocyte lipolysis.

Our study revealed several small but notable differences between Adipoq-AAKO and aP2-AAKO mouse models (18, 19), most notably in energy homeostasis and fat mass distribution (summarized in Supplemental Table 3). Specifically, rather than increasing body weight (18) and/or fat mass (18, 19), male Adipoq-AAKO mice maintained overall energy homeostasis and fat mass during the 24-week course of this study. Accordingly, their energy intake and expenditure were comparable in the ad libitum-fed state, even though overnight fasting resulted in reduced VO₂ in Adipoq-AAKO mice, consistent with previous results (19). Nevertheless, this result is quite remarkable given the comparable lipolytic defect and impairment in BAT morphology/function (18, 19). The absence of an overall increased total fat mass in Adipoq-AAKO mice is likely due to a larger decrease in liver mass and fat content. Furthermore, in contrast to increased WAT mass (18, 19) in aP2-AAKO mice, WAT depots were either unchanged or decreased in Adipoq-AAKO mice. Our data suggest that these results may be due to compensatory down-regulation of adipocyte lipid uptake, lipid synthesis, and/or adipogenesis, parameters that were either not evaluated (19) or not present under the conditions studied (chow-fed mice of unknown age) (18) in aP2-AAKO mice. Consistent with our data, Mottillo et al recently demonstrated that lipid oxidation and lipid synthesis are coupled during β₃-adrenergic-stimulated adipocyte lipolysis (29). Our data support this finding and further suggest that this coupling is also physiologically relevant during chronic inhibition of lipolysis. A previous study in aP2-AAKO mice has additionally implicated enhanced autophagy as another mechanisms that compensates for loss of ATGL (19), although this mechanism was not directly tested in our study. The absence of progressive fat mass expansion in our model is consistent with the phenotype of humans with ATGL deficiency (27, 28). Nevertheless, several other potential reasons might contribute to difference between the Adipoq- and aP2-AAKO models, including difference in ATGL deletion from non-AT cells, including immune cells, differences in the pattern of ATGL deletion in adipocytes, and/or differences in experimental conditions. Additional studies, directly comparing both models would be required to more specifically characterize relevant differences.

Despite the above differences, the Adipoq-Cre and aP2-Cre models share many similarities. Adipoq-AAKO mice have improved glucose homeostasis and insulin signaling in the liver, which is consistent with aP2-AAKO mice (18, 19). A recent study suggest that one potential mechanism for this effect is that adipocyte lipolysis increases hepatic acetyl coenzyme A levels, which in turn drive hepatic insulin resistance, an effect that is blocked in aP2-AAKO mice (35). Our study further suggests that this improvement in hepatic insulin sensitivity results, at least in part, from protection against hepatic steatosis and subsequent hepatic immune cell recruitment/activation and inflammation. Obesity-associated elevation of circulating NEFAs and ectopic lipid accumulation in metabolically relevant tissues such as liver are known to promote insulin resistance by multiple mechanisms, including activation of inflammatory pathways (1). In addition, DIO promotes hepatic accumulation of specific subpopulation of immune cells, particularly CD11c+ cells, as well as increases expression of proinflammatory cytokines (25). However, these effects are notably absent in Adipoq-AAKO mice, suggesting that adipocyte lipolysis is a key determinant of hepatic metabolic and immune responses to DIO. Furthermore, these responses are not fasting dependent, because they occur in ad libitum-fed mice. Of interest, both deletion (current study) and overexpression (36) of ATGL in adipocytes improves hepatic steatosis and insulin resistance, suggesting that adipocyte lipolysis may be a therapeutic target for these disorders. In contrast, modulating ATGL action within skeletal muscle (22) and liver (37–39) do not significantly influence glucose homeostasis and insulin signaling. Interestingly, despite a comparable reduction in adipocyte lipolysis, GAKO mice have increased rather than decreased susceptibility to hepatic inflammation, presumably because hepatic ATGL action is required not only for hepatic lipid homeostasis but for activation of PPARα's antiinflammatory effects (40). Thus, ATGL action in different tissues differentially affects both metabolic and immune phenotypes in the liver.

Our study also supports a role for ATGL-mediated adipocyte lipolysis in the immune responses within AT. Specifically, inhibition of adipocyte lipolysis ameliorates the acute inflammatory response within AT but not the chronic inflammatory response to DIO. Regarding acute lipolysis, several studies have shown that acute activation of lipolysis, either pharmacologically (ie, β₃-adrenergic stimulation) (6) or physiologically (ie, weight loss and fasting) (5), increases AT immune cell recruitment and activation. Likewise, global deletion of hormone-sensitive lipase (HSL) (41) or ATGL (5) reduce immune cell recruitment into AT in response to acute lipolytic stimuli. Our data demonstrating reduced recruitment of CD11c+ and F4/80+ cells to AT of AAKO mice after β₃-adrenergic stimulation corroborate these finding and further suggest that these effects are indeed due to inhibition of adipocyte lipolysis rather than to HSL or ATGL action within immune cells themselves. This immediate immune response to adipocyte lipolysis has been proposed to play a “physiological” role in regulating lipolysis by sequestering potentially deleterious lipids (5) and/or by promoting AT remodeling (42, 43). This process is still poorly understood, and the mechanisms by which adipocyte lipolysis might alternatively promote a physiological vs pathological immune response in AT remain controversial.

The impact of ATGL action on the chronic immune response in AT is even more complex. Unlike the reduction of immune cell infiltration observed in liver or with acute lipolysis, chronic adipocyte ATGL deletion tended to exacerbate rather than protect against AT immune cell infiltration and inflammation. This effect was quite dramatic in BAT but was also present in WAT where changes in morphology and function were less pronounced. These alterations in AT immune phenotype were present in both chow- and HFD-fed AAKO mice, suggesting that these changes are not entirely diet-dependent. Consistent with our findings in AAKO mice, HSL deficiency in both mice (3, 41, 44) and humans (45) is associated with increased AT inflammation in the basal/unstimulated state, despite comparable reduction in the immune response to acute lipolysis. Additional long-term studies are required to better characterize the natural history and physiological impact of the AT immune response to modulation of ATGL (vs HSL) action, including its effects on glucose homeostasis/insulin action and adipocyte function.

There are multiple mechanisms by which modulation of adipocyte ATGL action could influence the AT immune response. However, it is first worth noting that ATGL deficiency was not sufficient to protect against DIO-induced AT inflammation. These data suggest that factors other than regulated adipocyte lipolysis ultimately drive the AT inflammatory response to DIO - the most obvious mechanism being unregulated release of FAs from dead or dying adipocytes (3, 46). In support of this hypothesis, a recent study has shown that endogenous adipocyte-derived lipids trigger the AT inflammatory response (47), although the quantities and identities of the responsible lipid species as well as the precise mechanisms that contribute their release (via adipocyte death or otherwise) remain unknown. Given the increase in the AT immune response in AAKO mice, it is possible that chronic adipocyte ATGL deficiency contributes to adipocyte tissue dysfunction and/or adipocyte death, leading to either proinflammatory adipokine/cytokine production and/or unregulated release of FAs that enhance immune cell recruitment or activation. Alternatively, ATGL action may produce lipid ligands that directly or indirectly inhibit the inflammatory response (either cell autonomously within adipocyte themselves or by acting in nonadipocyte cells such as AT MCs). Indeed, accumulating evidence suggests that ATGL produces lipid ligands that regulate PPAR nuclear transcription factors (ie, PPARα) and their downstream targets (48–50). Both PPARα and PPARγ independently influence lipid metabolism and inflammation (14), and both are influenced by adipocyte ATGL deletion in this study. In support of this hypothesis, recent studies by Ciriolo's group have suggested that ATGL action protects against the AT immune response to nutritional and age-related stress by producing FAs that not only provide energy but also serve as ligands for PPARs (12, 13). They further show that GAKO mice have increased mRNA expression of inflammatory cytokines in AT (12, 13), similar to our AAKO mice. Additional studies are required to more systematically dissect the relative contribution of these mechanisms to the AT metabolic and immune responses.

In summary, this study clarifies the metabolic consequences of impaired ATGL-mediated adipocyte lipolysis and underscores the complex relationship between adipocyte lipolysis and the immune response. Specifically, reduced adipocyte lipolysis improves serum lipids and systemic glucose homeostasis, at least in part by reducing steatosis, immune cell infiltration, and insulin signaling in the liver. Nevertheless, adipocyte lipolysis also contributes to the immune response within AT, both after stimulated lipolysis and in the unstimulated basal state. Additional studies are required to more specifically define the natural history and (patho)physiological impact of altered adipocyte lipolysis on adipocyte biology, systemic metabolism, and the immune response. In addition, the relative contribution of ATGL action within adipocytes vs other cell types (ie, immune cells) to these outcomes remains to be determined. These studies will provide valuable information related to how this fundamental metabolic process might be manipulated for therapeutic benefit in the treatment of cardiometabolic disease.

Acknowledgments

We thank the next contributions: University of Pittsburgh Center for Biological Imaging for equipment and expertise related to imaging and University of Pittsburgh Center for Metabolic and Mitochondrial Medicine for expertise related to metabolism.

Author contributions: G.S. and E.E.K. were the project leaders and contributed to all aspects of this work and all other authors contributed to experimental design, execution of experiments, data analysis, and/or manuscript preparation. All authors contributed intellectually to this work and reviewed/edited the manuscript. E.E.K. 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.

Present address for M.T.S.: Department of Biology and Molecular Biology, Montclair State University, Montclair, NJ 07043.

Present address for G.S.: Institute of Molecular Biosciences, University of Graz, Graz, Austria 8010.

Present address for M.K.B.: Institute of Bioscience and Technology, Shri Ramswaroop Memorial University, Uttar Pradesh, India 225003.

Present address for C.F.Y.: Division of Endocrinology and Metabolism, Medical College of Virginia 23219.

This work was supported by National Institutes of Health (NIH) Grant R01DK090166, a Howard Hughes Medical Institute Physician-Scientist Early Career Award, and a University of Pittsburgh Department of Medicine Junior Scholar Award (E.E.K.); NIH Grant R01DK072162 (to R.M.O.); Erwin Schrödinger Fellowship J3221-B19 funded by the Austrian Science Fund (FWF) (to G.S.); NIH Grant K12HD063087 (to M.N.M.); NIH Grant T32DK007042 (to B.A.S.); and the Endocrine Fellows Foundation's Marilyn Fishman Grant for Diabetes Research (C.F.Y.).

Disclosure Summary: The authors have nothing to disclose.

Abbreviations

 
• AAKO

  adipocyte-specific ATGL knockout

 
• Adipoq

  adiponectin

 
• AT

  adipose tissue

 
• ATGL

  adipose triglyceride lipase

 
• ATM

  AT macrophage

 
• BAT

  brown AT

 
• CD

  cluster of differentiation

 
• DIO

  diet-induced obesity

 
• FA

  fatty acid

 
• GAKO

  global ATGL knockout

 
• H&E

  hematoxylin and eosin

 
• HFD

  high-fat diet

 
• HSL

  hormone-sensitive lipase

 
• IF

  immunofluorescence

 
• MC

  mononuclear cell

 
• MCP1

  monocyte chemoattractant protein-1

 
• NEFA

  non-esterified fatty acid

 
• PFA

  paraformaldehyde

 
• PPAR

  peroxisome proliferator-activated receptor

 
• SVC

  stromal vascular cell

 
• TAG

  triacylglycerol

 
• VO₂

  oxygen consumption

 
• WAT

  white AT.