Myeloid-specific estrogen receptor α deficiency impairs metabolic homeostasis and accelerates atherosclerotic lesion development

Abstract

ERα is expressed in macrophages and other immune cells known to exert dramatic effects on glucose homeostasis. We investigated the impact of ERα expression on macrophage function to determine whether hematopoietic or myeloid-specific ERα deletion manifests obesity-induced insulin resistance in mice. Indeed, altered plasma adipokine and cytokine levels, glucose intolerance, insulin resistance, and increased adipose tissue mass were observed in animals harboring a hematopoietic or myeloid-specific deletion of ERα. A similar obese phenotype and increased atherosclerotic lesion area was displayed in LDL receptor-KO mice transplanted with ERα^−/− bone marrow. In isolated macrophages, ERα was necessary for repression of inflammation, maintenance of oxidative metabolism, IL-4–mediated induction of alternative activation, full phagocytic capacity in response to LPS, and oxidized LDL-induced expression of ApoE and Abca1. Furthermore, we identified ERα as a direct regulator of macrophage transglutaminase 2 expression, a multifunctional atheroprotective enzyme. Our findings suggest that diminished ERα expression in hematopoietic/myeloid cells promotes aspects of the metabolic syndrome and accelerates atherosclerosis in female mice.

The metabolic syndrome reflects an aggregate of abnormalities including obesity, inflammation, hypertension, glucose intolerance, dyslipidemia, and insulin resistance (1–3), and the clustering of these abnormalities is associated with the increased risk of numerous chronic diseases including type 2 diabetes, atherosclerosis, and certain forms of cancer (4–7). Although the incidence of heart disease has plateaued in men over the past two decades, atherosclerosis and type 2 diabetes continue to escalate in women at an alarming rate (8–10). Obesity is viewed as a primary factor contributing to the rising rates of chronic disease in middle-age women (5); however, the therapeutic targets to combat obesity have yet to be identified.

Human inactivating mutations in the estrogen receptor (ER) α isotype recapitulate aspects of the metabolic syndrome (11–14), and diminished ERα expression is associated with obesity in women (15). Indeed, rodents harboring a homozygous Esr1 null mutation display glucose intolerance, heightened tissue inflammation, insulin resistance, and marked obesity (16–18). Furthermore, several gene clusters within the mouse atherosclerosis locus Ath11 acting in early lesion formation are ERα-regulated (19). These findings are consistent with the hypothesis that impaired ERα action is a contributing factor in the pathobiology of the metabolic syndrome and elevated disease risk.

Inflammation has emerged as a central underpinning in obesity-induced insulin resistance (20, 21), and classically activated immune cells contribute to proinflammatory signaling in glucoregulatory tissues (22–27). Although modulation of inflammatory signaling in macrophages by estradiol in large part is ERα-dependent (28–30), the impact of ERα expression on in vivo macrophage function and the contribution of this cell type to the development of the obesity-insulin resistance phenotype observed in whole body ERα^−/− mice is unknown.

Herein, we show that ERα is critical for the maintenance of macrophage metabolism, and its expression is similarly and markedly induced by IL-4 and estradiol. Furthermore, ERα is required for macrophage IL-4 responsiveness, including induction of key transcription factors, markers of alternative activation, and transglutaminase 2 expression, as well as maximal phagocytic capacity. In female mice, these defects in macrophage function resulting from ERα deletion manifest glucose intolerance, insulin resistance, tissue inflammation, obesity, and increased atherosclerotic lesion area.

Results

Hematopoietic Deletion of ERα.

ERα protein and mRNA expression levels were markedly and equally reduced in bone marrow-derived macrophages (BM-Mφ) from 10-wk-old obese female C57Bl6 Lep^ob mice compared with lean age-matched controls (Fig. S1 A and B). To investigate the impact of ERα expression on macrophage function and the contribution of this cell type in the pathogenesis of obesity and insulin resistance, two independent mouse lines with a hematopoietic or myeloid-specific deletion of ERα were generated. Bone marrow chimeras were produced by injecting irradiated WT recipient mice with marrow cells from WT or ERα^−/− mice (24), yielding WT animals with either WT (BMT-WT) or ERα knockout (BMT-KO) hematopoietic cells. To estimate homing and incorporation of bone marrow-derived cells into glucoregulatory tissue beds, we examined the presence of a bone marrow pseudomarker unique to the ERα^−/− mice (neomycin, Neo, resistance cassette inserted into exon 2 of the Esr1 gene) (17). In the bone marrow transplantation (BMT)-KO mice, we found that 40% of cells in adipose tissue, whereas only 20% and 15% of cells in liver and skeletal muscle, respectively, were of bone marrow origin (Fig. S1C).

A second line of mice with a myeloid-specific ERα deletion (MACER) were generated by crossing ERα floxed mice (31) with a transgenic line in which Cre recombinase was driven by the Lysozyme (Lys) M promoter (32). PCR analyses of genomic DNA showed recombination in BM-Mφ from MACER mice (Fig. S1D), consistent with previous reports that LysM Cre-mediated gene excision occurs almost exclusively in macrophages and neutrophils (22, 24, 32). In MACER mice, ERα protein and transcript levels were minimally detected in peritoneal (Fig. S1E) and BM-Mφ (Fig. S1F), but fully expressed in skeletal muscle, adipocytes and hepatocytes (Fig. S1F). Transcript levels of Esr2 (ERβ) and the transmembrane G protein coupled estrogen receptor (Gper, GPR30) were minimally expressed in BM-Mφ, and no compensatory increase in expression was observed in the absence of ERα (Fig. S1G), findings consistent with Murphy et al. (33) showing α as the predominant estrogen receptor in macrophages.

Myeloid-Specific ERα Deletion Causes Glucose Intolerance and Insulin Resistance.

Hematopoietic/myeloid-specific ERα deletion did not affect estradiol concentration, body mass (Tables S1–S3), ambulatory movement, feeding, or oxygen consumption rates (Fig. S2). Although fasting blood glucose was similar between the genotypes (Tables S1 and S2), insulin levels were markedly increased by twofold and 1.5-fold (P = 0.004 and P = 0.01) in BMT-KO and MACER mice, respectively. Adiponectin was reduced by 34% (P = 0.0006) in MACER vs. f/f (Table S2), whereas no change in plasma triglyceride or total cholesterol was detected between the genotypes (Table S2). Select plasma chemokines and cytokines were analyzed and of the 10 analytes measured, IFN-γ, IL-1β, IL-6, MCP-1, Rantes, and TNF-α were all elevated two- to fourfold in MACER vs. f/f Control; however, only Rantes reached statistical significance (Fig. S3; P = 0.03).

Glucose tolerance was significantly diminished in BMT-KO and MACER mice compared with BMT-WT and f/f Control mice (P = 0.043 and P = 0.026, respectively; Fig. 1 A and B). The rate of exogenous glucose infusion (GIR) required to maintain euglycemia during euglycemic-hyperinsulinemic clamp studies was significantly reduced, by 33% (P = 0.001) in BMT-KO and 50% (P = 0.0014) in MACER mice compared with respective controls (Tables S1 and S2). The insulin-stimulated glucose disposal rate (IS-GDR), reflecting skeletal muscle insulin sensitivity, was reduced by 24% in BMT-KO (P = 0.04; Fig. 1C) and 51% in MACER (P = 0.018; Fig. 1D) compared with BMT-WT and f/f. Furthermore, although insulin promoted a twofold increase in 2-deoxyglucose uptake into soleus muscle from f/f mice ex vivo, glucose uptake into MACER muscle was diminished by 50% (P = 0.032; Fig. 1E). A similar impairment in insulin-stimulated Akt phosphorylation in MACER muscle (Fig. 1F) was also observed.

Fig. 1.

Hematopoietic/myeloid-specific ERα deletion causes glucose intolerance and insulin resistance in female mice. (A and B) Glucose intolerance in normal chow (NC)-fed BMT-KO vs. BMT-WT and MACER vs. f/f Control mice (n = 8 per group; area under the glucose curve, AUC). (C and D) Reduced insulin-stimulated glucose disposal rate (IS-GDR) in BMT-KO and MACER mice vs. BMT-WT and f/f Control, respectively (n = 6–8 per group). (E) Impaired insulin-stimulated 2-deoxyglucose uptake into soleus muscle in MACER compared with f/f Control (n = 8 per genotype). (F and G) Akt total protein and phosphorylation in insulin-stimulated soleus muscle from f/f Control and MACER, and C2C12 myotubes pretreated with conditioned media from WT or ERαKO BM-Mφ (n = 6 per condition). (H) JNK 1/2 total protein and phosphorylation in quadriceps muscle from BMT-KO vs. BMT-WT (n = 6 per genotype). (I–K) Blunted suppression of HGP (HGP % suppression) by insulin and hepatic steatosis in MACER and BMT-KO compared with f/f Control and BMT-WT mice (n = 6–8 per group). (L and M) JNK1/2 and IKKα/β total protein and phosphorylation in liver samples from BMT-KO and BMT-WT (n = 6 per genotype). *P < 0.05, between genotype difference.

Given detection of bone marrow-derived cells in skeletal muscle of BMT-KO mice, we next investigated a potential role for direct paracrine action of macrophages on muscle. Insulin-stimulated Akt phosphorylation was significantly reduced in C2C12 myotubes pretreated with conditioned media (CM) from ERα-deficient macrophages (Fig. 1G). Although the insulin resistance producing factors secreted by classically activated macrophages have yet to be identified, release of substances promoting local tissue inflammation are likely candidates. Indeed, our findings support this notion because JNK phosphorylation was elevated 96% in quadriceps muscle from BMT-KO versus BMT-WT (Fig. 1H).

Hepatic insulin resistance can also explain in part the reduction in insulin sensitivity seen in BMT-KO and MACER mice. Suppression of hepatic glucose production (HGP) was significantly blunted (P = 0.037; Fig. 1I) in MACER mice compared with f/f Control animals, and this hepatic insulin resistance was associated with increased liver fat content shown by MRI analysis (66%↑; P = 0.01; Fig. 1J) and by histology (Fig. S4A). Hepatic lipid content was highly associated with body weight in MACER mice, suggesting a strong influence of adiposity on liver function (Fig. S4B). Hepatic insulin resistance was even more severe in BMT-KO mice (Fig. 1K), and similar to skeletal muscle, heightened inflammatory signaling was detected in liver samples from these animals (Fig. 1 L and M).

Macrophage ERα Deletion Causes Increased Adiposity and Atherosclerotic Lesion Size.

Similar to findings in whole body ERα^−/− mice (18), plasma leptin and PAI-1 levels, as well as periovarian and total adipose tissue mass, were markedly increased in MACER and BMT-KO compared with respective controls (Fig. 2 A–D). Increased expression of chemokines and markers of immune cell infiltration and inflammation (Fig. 2E) were paralleled by increased adipocyte size and abundance of F4/80 positive crown-like structures in MACER perigonadal adipose tissue (Fig. 2F). The close proximity of macrophages to adipocytes and the increased presence of bone marrow-derived cells in MACER gonadal fat support a paracrine cross-talk between the two cell types, and conditioned media (CM) from ERα-deficient cultured BM-Mφ attenuated insulin-induced glucose uptake (Fig. 2G) and Akt phosphorylation (Fig. 2H) in 3T3L1 adipocytes. These data demonstrate that hematopoietic/myeloid-specific ERα deletion promotes inflammation, insulin resistance, and obesigenic effects in mice.

Fig. 2.

Altered circulating adipokines, obesity, and increased atherosclerotic lesion area in female mice with a hematopoietic/myeloid-specific ERα deletion. (A and B) Plasma leptin and PAI-1 levels, (C) Gonadal adipose tissue mass in f/f Control, MACER, BMT-WT, and BMT-KO. (D) Total fat mass for f/f Control vs. MACER. (E) Emr1, CD68, Cd3e, Ccl5, and Ifng expression (n = 6–12 animals per genotype). (F) H&E (10×), F4/80 staining (40×), and adipocyte size in gonadal adipose tissue from NC-fed MACER vs. f/f Control mice (size marker = 100 μm). (G) Insulin-stimulated 2-deoxyglucose uptake and Akt phosphorylation in adipocytes after preconditioning with media from ERα WT or KO BM-Mφ (H) (n = 6). (I) Sudan IV staining of aortas harvested from BMT-WT and BMT-KO mice on LDLR-KO background (n = 12 mice per group). *P < 0.05, between genotype difference.

To investigate the impact of macrophage ERα expression in the pathobiology of atherosclerosis, we performed en face analyses on aortas harvested from LDL receptor (LDLR)-KO recipient mice of WT and ERα^−/− bone marrow. In addition to increased adipose mass (Fig. 2C), and elevated leptin (Fig. 2A) and PAI-1 (Fig. 2B) levels, BMT-KO/LDLR-KO mice also showed a twofold increase (P = 0.02) in atherosclerotic lesion area (Fig. 2I). These differences could not be attributed to glycemia or total plasma cholesterol as concentrations were identical between the genotypes (Table S3). To investigate the underpinnings of accelerated atherosclerotic lesion development in BMT-KO/LDLR-KO, we further assessed the role of ERα in the regulation of macrophage functionality. Phagocytic capacity, an essential antiatherosclerotic action of macrophages, was markedly impaired in LPS- (Fig. 3A) and LDL-treated (Fig. 3B) BM-Mφ from ERαKO vs. WT mice. Furthermore, basal cholesterol uptake was increased by 28% (P = 0.001; Fig. 3C) and cholesterol efflux to HDL particles was reduced by 21% (P = 0.0005; Fig. 3D) in KO vs. WT macrophages supporting foam cell formation. Indeed, OxLDL induction of ApoE was significantly blunted in KO macrophages (Fig. S5), and ERα, in part, is a likely mechanistic explanation for increased atherosclerotic lesion area in BMT-KO/LDLR-KO vs. BMT-WT/LDLR-KO mice.

Fig. 3.

Impaired macrophage function in female mice with hematopoietic/myeloid-specific ERα deletion. Impaired LPS- (A) and OxLDL-induced (B) phagocytosis in BM-Mφ from MACER and BMT-KO vs. WT (n = 4–6 per genotype). (C) Increased basal cholesterol uptake in BM-Mφ from KO vs. WT. (D) Impaired cholesterol efflux to HDL particles in BM-Mφ from KO vs. WT. (E) Estradiol-induced Apoe, Abca1, and Tgm2 expression (relative to basal vehicle-treated) in KO vs. WT BM-Mφ (n = 4–6 per genotype). (F Upper) Schematic of full-length and truncated, 2-kb (ERE replete) and 1-kb (ERE deleted), Tgm2 promoters. (F Lower) Tgm2 promoter activation by estradiol is blunted in the 1-kb promoter lacking the consensus ERE (n = 3 per condition). (G) Estradiol and IL-4–induced Esr1 expression in WT BM-Mφ (n = 6 per condition). (H) Estradiol, IL-4, and combined estradiol + IL-4 induction of IL-4rα expression in WT and KO BM-Mφ (n = 6 per condition). (I) Reduced IL-4 receptor protein and IL-4–induced STAT6 phosphorylation in KO vs. WT BM-Mφ. (J) IL-4–induced expression of transcription factors (Ppard, Pparg, and Ppargc1; arbitrary units, AU), markers of alternative activation (Arg1, Chi3l3, Retnla, and Tgfb1). (K) Tgm2 (expression above basal) in KO vs. WT BM-Mφ assessed by q-PCR (n = 6 per genotype, normalized to 1.0). (L) FACS analysis of cell surface marker fluorescence for MHCII and CD11c in adipose tissue F4/80+ Mφ from vehicle and IL-4–treated f/f Control vs. MACER mice (n = 3 observations per genotype and treatment condition). (M) C¹⁴-labeled palmitate oxidation for KO vs. WT BM-Mφ (n = 6 per genotype). (N–P) Palmitate-induced Ifng, IL-6, and IL-1β gene expression in KO vs. WT BM-Mφ assessed by quantitative PCR (n = 6 per genotype). *P < 0.05, between genotype difference; ¹P < 0.05, difference between treatment condition, within genotype.

Previous reports suggest that transglutaminase 2 (Tgm2) is an important enzyme regulating reverse cholesterol transport, wound healing, apoptotic cell phagocytosis and, thus, the severity of lesion development (34). We identified Tgm2, ApoE, and Abca1 as estrogen-responsive genes whose expression was reduced in both KO BM-Mφ and peritoneal macrophages from MACER mice (Fig. 3E and Fig. S6). Indeed, promoter regions from all three genes contain potential ERα binding sites (Fig. 3F and Fig. S7). Given its multifunctional, antiatherosclerotic actions, and the presence of a full consensus estrogen response element (ERE) in its 11-kb promoter, we investigated the role ERα in the induction Tgm2 expression. Although estradiol induced robust activation of the 2-kb ERE-containing promoter comparable with the control ERE-luciferase plasmid, limited stimulation by estradiol above vehicle treatment was observed for the 1-kb Tgm2 promoter lacking the putative ERα binding site (Fig. 3F, Lower). Collectively, these findings suggest that ERα is critical for the basal expression and ligand-mediated induction of several macrophage-specific genes with well-established links to atherosclerotic lesion formation.

ERα-Deficient Macrophages Are Refractory to IL-4–Induced Alternative Activation.

Alternative macrophage activation is associated with antiinflammatory actions, wound healing, and a lean mouse phenotype (25, 35, 36). We found that ERα expression was elevated in WT BM-Mφ in response to estradiol and IL-4 stimulation (Fig. 3G). Surprisingly, estradiol was equally as effective as IL-4 to induce IL-4 receptor expression (Fig. 3H); however, these effects were blocked in ERαKO cells. Moreover, basal IL-4 receptor protein levels and IL-4–induced p-STAT6 were diminished in ERαKO vs. WT BM-Mφ (Fig. 3I). IL-4–induced expression of key transcription factors, Pparg, Ppard, and Ppargc1b (Fig. 3J), and standard markers of alternative activation (Chi3L3, Retnla, and Tgfb1; Fig. 3J) and Tgm2 (Fig. 3K), were blunted in KO BM-Mφ. Unexpectedly, no difference in Arginase expression was detected (Fig. 3J). Similar to observations for BM-Mφ, adipose tissue macrophages obtained from MACER mice were refractory to the suppressive effects of IL-4 on cell surface markers of inflammation compared with f/f Control (Fig. 3L and Fig. S8). Consistent with previous findings by Vats et al. (35) linking inflammation with cellular oxidative metabolism, we observed a 47% (P = 0.002) reduction in ¹⁴C-labeled palmitate oxidation in BM-Mφ lacking ERα (Fig. 3M). Furthermore, palmitate treatment increased expression of Ifng, IL6, and IL1b in ERαKO BM-Mφ vs. WT (Fig. 3 N–P). Our findings indicate that ERα is critical for the maintenance of macrophage function including oxidative metabolism and responsiveness to LPS, IL-4, estradiol, and proinflammatory fatty acids.

Discussion

Although strong clinical and experimental evidence supports an important role for ERα in the regulation of glucose homeostasis and adipose tissue development, the cell type(s) responsible for conferring the obesity, insulin resistance phenotypes observed in humans with inactivating receptor mutations and in mice harboring a homozygous null mutation remain unknown. Considering the now well-established role for macrophages in the control of insulin sensitivity and adiposity, we set out to determine which aspects of the metabolic syndrome could be recapitulated in mice lacking ERα in hematopoietic/myeloid cells.

Macrophages devoid of ERα were unable to respond appropriately to prototypical cues including IL-4, LPS, saturated fatty acids, and oxidized LDL. Thus, the professional skill set unique to this cell type (e.g., phagocytosis, wound healing, secretory repertoire) was impaired or garnished in the context of ERα deficiency and these aberrant cellular processes promoted aspects of the metabolic syndrome (e.g., glucose intolerance, insulin resistance, and obesity), as well as atherosclerotic lesion development in mice (Fig. S9). Notably, insulin, leptin, and PAI-1 levels were increased in all three mouse models lacking ERα in hematopoietic/myeloid cells, and this circulating factor profile is often observed in obese humans with atherosclerosis (37–39).

Herein, we made the observation that IL-4 treatment induced ERα expression in WT cells, concomitant with increased expression of PPAR-γ, PPAR-δ, and PGC1-β. Notably, IL-4–induced expression of these factors was markedly reduced, and IL-4–mediated suppression of MHCII and CD11c was blunted in ERα-deficient macrophages. In addition, IL-4 receptor α protein levels, IL-4–induced activation of STAT6, and expression of canonical markers of alternative activation were also diminished in ERαKO cells. A direct effect of ERα on Il4r expression is supported by the presence of an ERα tethering site and partial ERE in the IL-4r promoter. Furthermore, PPAR expression is reduced in response to ERα deletion from other glucoregulatory cell types, implicating a conserved transcriptional program under ERα control (18). Therefore, these data suggest that ERα is critical in the regulation of alternative activation and could modulate IL-4 action via transcription factor cross-talk and or upstream effects on IL-4 receptor signal transduction.

In addition to obesity and whole body insulin resistance, atherosclerotic lesion area was increased twofold in BMT-KO mice. This observation is consistent with previous findings by Hodgin et al. (40) showing advanced aortic lesion formation in female ERα/ApoE double knockout mice. Herein, we made the observation that estradiol and IL-4 induce the expression of transglutaminase (Tgm) 2, an enzyme protective against atherosclerotic lesion development (34). Induction of Tgm2 by these stimuli was blunted in macrophages devoid of ERα, and we identified a single consensus ERE in the 11-kb upstream promoter essential for activation by estradiol. Considering that hematopoietic-specific Tgm2 KO animals display increased atherosclerosis (34), it is reasonable to assume that reduced Tgm2 expression contributes in part to the increased lesion area in aortas from BMT-KO/LDLR-KO mice. Similar to Tgm2, macrophage ApoE (41, 42) and Abca1 (43, 44) also exert important atheroprotective actions, and in the case of ApoE, expression is augmented by estradiol in an ERα-dependent fashion (45, 46). Importantly, a similar reduction in Abca1 and ApoE expression was observed in ERαKO BM-Mφ in culture and in peritoneal macrophages obtained from MACER mice. Expression levels of both ApoE and Abca1 are known to be reduced by inflammatory cytokines IFN-γ and IL-1β, and positively regulated by TGF-β (47–49) and OxLDL (50). We found that induction of ApoE by OxLDL was markedly blunted in macrophages lacking ERα; thus, reduced ApoE and Abca1 expression in KO macrophages could have resulted from both direct effects of ERα deficiency and indirect action of heightened cellular inflammation and reduced TGF-β.

In line with findings from Odegaard (25) and Vats et al. (35) showing elevated oxidative metabolism in alternatively vs. classically activated macrophages, fatty acid oxidation rates were markedly reduced and inflammatory gene expression elevated (Ifng, IL6, and IL1b) by ERα deficiency. Furthermore, ERαKO macrophages secreted proinflammatory factors promoting insulin resistance in glucoregulatory cell types in culture. Identification of insulin resistance producing factors secreted by ERα-deficient macrophages are necessary as these secretory products are likely proinflammatory and contribute to the heightened incidence of metabolic-related disease observed in postmenopausal women (5).

The current studies are of clinical and therapeutic interest because our findings show that impaired ERα action in macrophages is causal for the development of aspects of the metabolic syndrome and increased atherosclerotic lesion formation in female mice. Whether reduced or dysfunctional ERα, independent of circulating estradiol levels, is causal for chronic disease in pre- and postmenopausal women remains unknown. However, a strong association between inactivating mutations and polymorphisms in ERα with obesity, glucose intolerance, inflammation, and hypertension is observed in human subjects (12–14). Furthermore, ERα expression is reduced in adipose tissue from obese premenopausal women (15), and we find similar reductions in ERα expression in adipose tissue and bone marrow-derived macrophages from young female Lep^ob mice. In addition, we also show that macrophages devoid of ERα are refractory to the atheroprotective effects of estradiol and IL-4. Thus, it is conceivable that alteration in ERα function could explain, in part, heightened prevalence of metabolic syndrome in young women (5, 9, 51) and, possibly, the lack of anticipated atheroprotective benefit observed during recent hormone replacement trials. Followup studies investigating estrogen receptor status in pre- and postmenopausal women and women undergoing ERα-targeted adjuvant therapies to combat breast cancer are warranted.

Methods

Experimental methods are described in greater detail in SI Methods.

Animals.

Lean and obese Lep^ob C57Bl6 female mice were purchased from The Jackson Laboratory at 10–12 wk of age. Female flox/flox (f/f; n = 30) (31) were crossed into a transgenic line where Cre recombinase driven by the lysozyme M promoter (32) was used to generate myeloid-specific ERαKO mice (MACER; n = 30). BMTs were performed as described (24) in WT C57Bl6 and LDLR-KO mice by using bone marrow from WT C57Bl6 mice or C57Bl6 background ERα^−/− mice. All procedures were performed in accordance with the Guide for Care and Use of Laboratory Animals of the National Institutes of Health, and approved by the Animal Research Committee of the University of California, Los Angeles (UCLA).

Metabolic Cages.

Indirect calorimetry, feeding, and movement patterns were assessed over a 48-h period after 24-h chamber acclimation as described (18).

Circulating Factors and Glucose Tolerance Testing.

Plasma from fasted animals (≈22–30 wk of age) was analyzed for circulating factors: insulin, leptin, resistin, PAI-1, IFN-γ, IL-10, IL-1β, IL-4, IL-6, KC, MCP1, MIP1α, Rantes, TNF-α (Lincoplex; Millipore), adiponectin (RIA; Millipore), triglycerides (WAKO), cholesterol (WAKO), and estradiol (Siemens Diagnostics). After a 2-wk recovery, animals were fasted again and glucose tolerance tests performed as described (24).

Whole Body and Liver Composition.

Body composition and isolated liver fat content were determined by magnetic resonance imaging (MRI) on an EchoMRI 3-in-1 Body Composition Analyzer.

Hyperinsulinemic-Euglycemic Clamp Studies.

Glucose clamp studies were performed in chronically cannulated mice 3 d after surgery as described (24).

Cellular and Ex Vivo Skeletal Muscle 2-Deoxyglucose Uptake.

Adipocyte (3T3-L1) glucose uptake was performed in 12-well culture plates and estimated by using the 2-deoxy glucose method described (24, 52). Insulin-stimulated (60 μU/mL) glucose uptake into soleus muscle was assessed by using 2-deoxy glucose as described (53).

Macrophage Cholesterol Uptake and Efflux.

Macrophages were loaded with [1α, 2α(n)-³H]-cholesterol (1 μCi/mL; 250 μL of volume per well) in growth medium for 48 h and cholesterol efflux assessed as described (54).

Tgm2 Promoter Activation.

The Tgm2 promoter fragments were generated by PCR from BAC clone RP23-396G1 (Children's Hospital Oakland Research Institute).

Immunoblot Analysis.

Skeletal muscle, adipose tissue, and liver samples were treated by using methods described in greater detail in SI Methods.

Histology and Immunohistochemistry.

Perigonadal adipose tissue immunohistochemistry was performed by the University of California at San Diego (UCSD), Moore's Cancer Center, Histology and Immunohistochemistry Shared Resource Facility (24), and atherosclerotic lesion area in LDLR-KO mice was performed by the UCSD-UCLA DERC Mouse Phenotyping Core (55).

Stromal Vascular Cell Isolation and Flow Cytometry Analysis.

Stromal vascular cell isolation was carried out as described (23, 24). Flow cytometry analyses were performed by the Department of Pathology and the Jonsson Comprehensive Cancer Center Shared Resource Facility.

Quantitative PCR.

Mouse tissues were first homogenized using TRIzol reagent, and RNA was isolated and further cleaned by using RNeasy columns (Qiagen) with DNase digestion.