Lipid-Overloaded Enlarged Adipocytes Provoke Insulin Resistance Independent of Inflammation

Abstract

In obesity, adipocyte hypertrophy and proinflammatory responses are closely associated with the development of insulin resistance in adipose tissue. However, it is largely unknown whether adipocyte hypertrophy per se might be sufficient to provoke insulin resistance in obese adipose tissue. Here, we demonstrate that lipid-overloaded hypertrophic adipocytes are insulin resistant independent of adipocyte inflammation. Treatment with saturated or monounsaturated fatty acids resulted in adipocyte hypertrophy, but proinflammatory responses were observed only in adipocytes treated with saturated fatty acids. Regardless of adipocyte inflammation, hypertrophic adipocytes with large and unilocular lipid droplets exhibited impaired insulin-dependent glucose uptake, associated with defects in GLUT4 trafficking to the plasma membrane. Moreover, Toll-like receptor 4 mutant mice (C3H/HeJ) with high-fat-diet-induced obesity were not protected against insulin resistance, although they were resistant to adipose tissue inflammation. Together, our in vitro and in vivo data suggest that adipocyte hypertrophy alone may be crucial in causing insulin resistance in obesity.

INTRODUCTION

Adipose tissue is a key energy storage organ that regulates whole-body energy homeostasis. Its parenchymal function is storing excess energy in the form of triglycerides and converting them into free fatty acids and glycerol to provide energy upon demand. In addition, as an endocrine organ, adipose tissue synthesizes various adipokines that affect food intake, insulin sensitivity, and immune responses (1, 2). In obesity, adipose tissue expands as a result of increases in adipocyte size (hypertrophy) and adipocyte number (hyperplasia) and actively modulates the population of immune cells (3, 4).

Obese subjects exhibit proinflammatory responses, endoplasmic reticulum (ER) stress, hypoxia, and/or mitochondrial defects as a result of unbalanced energy inputs in adipose tissue, leading to consequent systemic insulin resistance (5–7). In obese adipose tissue, chronic and low-grade inflammation have been implicated in insulin resistance, with elevated F4/80⁺, CD11b⁺, and CD11c⁺ M1-like macrophages (8, 9). M1-like macrophages secrete various cytokines that impair insulin sensitivity through the induction of proinflammatory signaling, including activation of Jun N-terminal protein kinase (JNK) or NF-κB. Furthermore, in several animal models with genetic ablation of proinflammatory responses, such as JNK knockout mice and IκB kinase β (IKK-β) heterozygous mice, the animals are resistant to diet-induced obesity and/or insulin resistance because of decreased inflammation (10–13).

Since most metabolic complication studies have been performed using severe and relatively late stage obesity models, it is not fully understood which factors or cell types in adipose tissues are primarily responsible for insulin resistance at the early stage of obesity. Recently, we and others have reported that body weight, adipose tissue mass, adipocyte hypertrophy, adipose tissue inflammation, and insulin resistance would increase in mice fed a short-term (less than 1 week) high-fat diet (HFD) (14–18). Interestingly, depletion of macrophages or lymphocytes by clodronate treatment or Rag1 knockout mice did not attenuate insulin resistance in early obesity (16). These data imply that not only adipose tissue immune cells but also adipocyte changes, including adipocyte hypertrophy, may play a key role in the initiation of insulin resistance during early obesity. In addition, intima media thickness (IMT), which is a prediction marker of cardiovascular events and strongly associated with insulin resistance (19), was increased with severe obesity but was not influenced by the degree of systemic inflammation or adipose tissue macrophage accumulation (20). Very recently, it has been reported that early B cell factor 1 (EBF1) reduction caused adipocyte hypertrophy and insulin resistance but did not influence the inflammatory pathways in both mouse and human adipocytes (21). These emerging pieces of evidence have suggested that adipose tissue inflammation, which is one of the major factors of insulin resistance in severe obesity, might be dissociated from adipocyte hypertrophy-linked insulin resistance under certain conditions of obesity. Although various studies have suggested a close relationship between adipocyte hypertrophy, insulin resistance, and inflammation in obesity (22, 23), it is largely unclear whether enlarged adipocytes per se would initiate insulin resistance regardless of inflammation in obesity.

In this study, we have developed an adipocyte hypertrophy model with or without inflammatory responses and investigated the effects of adipocyte hypertrophy on insulin resistance. Collectively, our data suggest that adipocyte hypertrophy is sufficient to provoke insulin resistance, independent of a proinflammatory response, in early obesity.

MATERIALS AND METHODS

Animals and treatments.

Seven-week-old male C3H/HeN, C3H/HeJ, db/+, db/db, and C57BL/6J mice were obtained from Central Lab Animal Inc. (Seoul, South Korea). All mice were maintained under specific-pathogen-free conditions and were housed in solid-bottom cages with wood shavings for bedding in a room maintained at 25°C with a 12/12-h light/dark cycle (lights on at 07:00). After a stabilization period of at least 1 week, the mice (8 weeks old) were fed a normal chow diet (NCD) until they were fed a 60% HFD for the indicated times (Research Diets, Inc., NJ). The HFD mice were compared with age-matched chow-fed mice. The average initial body weights in each group of mice were not different. For the oral glucose tolerance test, the mice were fasted for 6 h, basal blood samples were taken, and glucose was injected orally (2 g/kg of body weight). Blood samples were drawn at 15, 30, 45, 60, 90, and 120 min after injection. For treatment with rosiglitazone, 12-week-old db/+ mice and db/db mice were injected with rosiglitazone (oral gavage, 15 mg/kg/day; Sigma-Aldrich, MO) for 1 month. All animal procedures were in accordance with the research guidelines for the use of laboratory animals of Seoul National University.

Cell culture.

3T3-L1 preadipocytes were grown to confluence in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% bovine calf serum (BCS; Invitrogen Life Technologies, Carlsbad, CA). Two days after the cells reached confluence (day 0), differentiation of the 3T3-L1 cells was induced in DMEM containing 10% fetal bovine serum (FBS; Invitrogen Life Technologies, Carlsbad, CA), methylisobutylxanthine (520 μM), dexamethasone (1 μM), and insulin (167 nM) for 48 h. The culture medium was replaced on alternate days with DMEM containing 10% FBS and 167 nM insulin.

FFA treatment.

Free fatty acids (FFAs; Sigma-Aldrich, MO) were conjugated with FFA-free bovine serum albumin (BSA) for administration to cells. Briefly, FFAs were dissolved in ethanol and diluted in DMEM containing 1% FBS and 2% (wt/vol) BSA for 10 min at 55°C. BSA-conjugated FFA-containing media were used to challenge cells.

Oil red O staining.

3T3-L1 adipocytes were washed twice with phosphate-buffered saline (PBS), fixed for 1 h with 3.7% formaldehyde in PBS, and subsequently dehydrated with 100% propylene glycol (Amresco, OH) for 5 min. After removing the propylene glycol, Oil Red O dye was added to the plate, which was incubated overnight. Subsequently, Oil Red O was removed and 85% propylene glycol was added to the plate, which was allowed to stand for 5 min. Finally, excess dye was washed away with distilled water until the background was clear. Images were obtained using an Evos Original microscope (Advanced Microscopy Group).

SEM.

3T3-L1 adipocytes were washed three times with Dulbecco's phosphate-buffered saline (DPBS) and fixed for 30 min with 5% glutaraldehyde in DPBS at 4°C. After removal of the glutaraldehyde, 1% osmium tetroxide in DPBS was added for 1 h at 4°C. Subsequently, osmium tetroxide was removed and washed thoroughly with DPBS. For dehydration, 30%, 50%, 70%, 90%, and 100% ethanols were serially added for 10 min each. After dehydration, samples were dried for 10 h at room temperature. Images were obtained using a field emission scanning electron microscope (FE-SEM; JEOL JSM-6700F).

Quantitative RT-PCR.

Total RNA was isolated from 3T3-L1 adipocytes and epididymal adipose tissues. cDNA was synthesized using the Moloney murine leukemia virus (M-MuLV) reverse transcriptase kit according to the manufacturer's instructions (Thermo Fisher Scientific, MA). The primers used for quantitative real-time (RT)-PCR were obtained from Bioneer (South Korea), and their sequences are provided in Table S1 in the supplemental material.

Western blot analysis.

3T3-L1 adipocytes were stimulated with or without 10 nM insulin for 20 min at 37°C. 3T3-L1 adipocytes were lysed with NETN buffer (20 mM Tris [pH 7.9], 1 mM EDTA, 100 mM NaCl, 0.5% NP-40, 1 mM Na₃VO₄, 100 mM NaF, and protease inhibitor cocktail tablets [Roche Diagnostics]). Total cell lysates were centrifuged at 12,000 rpm at 4°C for 15 min to remove fat debris. The protein concentration was determined using a bicinchoninic acid (BCA) assay kit (Pierce). Western blot analyses were conducted according to the manufacturer's protocol (Cell Signaling Technology). IκBα, IRS1, Akt/PKB, and phospho-Akt/PKB antibodies were purchased from Cell Signaling Technology; glycogen synthase kinse 3β (GSK3β) and phospho-GSK3β antibodies were from Transduction Laboratories; the phospho-IRS1 antibody was from Biosource; p65 antibody was from Santa Cruz Biotechnology; lamin B was from AbFrontier; and GAPDH (glyceraldehyde-3-phosphate dehydrogenase) antibody was from BD Biosciences.

THP-1 migration assays.

The migration of THP-1 human monocytes was measured using Transwell plates (Corning) with a pore size of 8 μm. THP-1 monocytes (10⁵ cells/well) were loaded onto the top plates with adipocyte-conditioned medium (ACM). The cells on the top or bottom of the insert were scraped and counted using a hematocytometer. The migration rate was calculated as the percentage of cells in the bottom versus the total number of cells. For preparation of ACM, 3T3-L1 adipocytes treated with long-chain fatty acids for 3 and 6 days were incubated in DMEM supplemented with 0.2% BSA for 15 h.

Glucose uptake assay.

3T3-L1 adipocytes were incubated in low-glucose DMEM containing 0.1% BSA for 16 h at 37°C. Cells were stimulated with or without 100 nM insulin for 20 min at 37°C. Glucose uptake was initiated by the addition of [¹⁴C]deoxyglucose at a final concentration of 3 mol/liter for 10 min in HEPES-buffered saline (140 mM NaCl, 5 mM KCl, 2.5 mM MgCl₂, 1 mM CaCl₂, and 20 mM HEPES [pH 7.4]). The reaction was terminated by separating the cells from HEPES-buffered saline and [¹⁴C]deoxyglucose. After washes in ice-cold PBS, the cells were extracted using 0.1% SDS, and scintillation counting was used to measure ¹⁴C radioactivity. The protein concentration was determined, and the radioactivity was normalized to the protein concentration.

In vitro glucose bioprobe uptake assay.

3T3-L1 adipocytes were cultured on an 8-well chamber plate (Lab-Tek II). After differentiation, the cell culture medium was changed to a low-glucose medium (without FBS), and the cells were maintained in the new medium for 4 h. The cells were then incubated for 1 h in glucose-deficient DMEM (without FBS). For continuous monitoring of cellular glucose uptake with a DeltaVision imaging system (GE Healthcare), the 8-well chamber plate was loaded on the stage of the microscope. After pretreatment with 5 μm GB-Cy3 (24), 100 nM insulin was administered. Fluorescence images were recorded every 2 min. The images were digitized and saved on a computer for further analysis. The temperature of the chamber was maintained at 37°C.

Ex vivo glucose bioprobe uptake assay.

Epididymal adipose tissues were removed and sliced into sections of 5 by 5 by 2 mm. Sliced samples were incubated in low-glucose DMEM containing 0.1% BSA for 30 min at 37°C. Sliced samples were incubated with 50 μM GB-Cy3 for 30 min in the presence or absence of 1 μM insulin. After several washes with PBS-Tween 20, the adipocytes were stained with fluorescein isothiocyanate-conjugated boron-dipyrromethene (BODIPY). The samples were then stained with a Vectashield solution (Vector Laboratories Inc.) containing 4′,6-diamidino-2-phenylindole (DAPI) and observed using a Zeiss LSM 700 confocal microscope (Carl Zeiss).

Lentivirus packaging and infection.

The lentiviral shuttle vectors containing the mCherry-GLUT4-Myc construct and three helper plasmids (encoding HIV-1 Gag-Pol, HIV-1 Rev, and vesicular stomatitis virus protein G [VSV-G] envelope protein) were cotransfected into HEK-293T cells according to established procedures. The medium was removed and replaced with fresh medium 12 h after transfection. The medium containing the viral particles was harvested 48 h later and filtered through 0.45-μm filters. The virus-containing medium along with 8 mg/ml of Polybrene (Sigma-Aldrich) was added to 3T3-L1 cells for overnight infection, and the medium was replaced with fresh 10% BCS–DMEM the following day.

TIRFM.

Total internal reflection fluorescence microscopy (TIRFM) imaging was performed using an inverted microscope system equipped with a 100× 1.45-numerical-aperture (NA) objective (Nikon). Images were collected using the NIS-Elements AR software. All experiments were performed at room temperature (22 to 25°C).

Statistical analysis.

Results represent data from multiple (three or more) independent experiments. Error bars represent standard deviations, and P values were calculated using Student's t test or analysis of variance (ANOVA).

Microarray data accession number.

Microarray data are available under Gene Expression Omnibus (GEO) accession number GSE65557 (http://www.ncbi.nlm.nih.gov/geo).

RESULTS

Treatment with long-chain fatty acids induces adipocyte hypertrophy.

To characterize hypertrophic adipocytes, we developed a cell culture model of adipocyte hypertrophy. Differentiated 3T3-L1 adipocytes were challenged for 6 days with saturated fatty acids (SFAs) (palmitic acid, C_16:0, and stearic acid, C_18:0) and a monounsaturated fatty acid (MUFA) (oleic acid, C_18:1) that are abundantly present in the diet (25, 26). Long-chain fatty acid-treated adipocytes became enlarged in time- and dose-dependent manners (Fig. 1A). Oil red O staining and scanning electron microscopy analysis revealed that lipid-overloaded hypertrophic adipocytes contained enlarged uniloculus-like lipid droplets (Fig. 1B and C). When adipocytes were treated with long-chain fatty acids, the percentage of adipocytes with large lipid droplets (>500 μm³) increased (Fig. 1D).

FIG 1.

Long-chain fatty acids induce adipocyte hypertrophy. 3T3-L1 adipocytes were differentiated and then cultured for another 6 days with various long-chain fatty acids at the indicated doses (250, 500, and 750 μM). CTL, control (BSA); PA, palmitic acid; SA, stearic acid; OA, oleic acid. (A) Microscopic images were obtained on the indicated days. (B) Oil red O staining of 3T3-L1 adipocytes treated with various fatty acids. (C) Scanning electron microscopy images of OA (500 μM)-treated hypertrophic adipocytes. (D) Nile red staining of lipid-overloaded 3T3-L1 adipocytes (top). The cells were fixed and stained with Nile red to visualize lipid droplets on day 6 after fatty acid treatment, and images were acquired using a confocal microscope. Lipid droplet (LD) volume distribution in lipid-overloaded 3T3-L1 adipocytes (bottom). LD volumes were measured from three-dimensional (3D) reconstructed images using Carl Zeiss ZEN. Scale bars, 20 μm.

MUFA-challenged hypertrophic adipocytes lack inflammatory responses.

Given that saturated fatty acids upregulate proinflammatory pathways (27, 28), we asked whether SFA-treated or MUFA-treated adipocytes would exhibit different inflammatory responses. Similar to tumor necrosis factor alpha (TNF-α)-treated adipocytes, SFA-induced hypertrophic adipocytes showed increased nuclear NF-κB (p65) and JNK phosphorylation and decreased cytosolic IκBα (Fig. 2A). However, MUFA-induced hypertrophic adipocytes did not significantly alter proinflammatory signaling cascades, unlike SFA-induced hypertrophic adipocytes. As shown in Fig. 2B, the mRNA levels of proinflammatory cytokines such as interleukin-6 (IL-6), serum amyloid A (SAA), TNF-α, and RANTES were elevated in SFA-treated adipocytes but not in MUFA-treated adipocytes. The mRNA levels of other genes such as adipogenic, lipogenic, lipolytic, glucose transporter, fatty acid transporter, ER stress, and hypoxic stress marker genes, were not significantly altered in SFA- and MUFA-overloaded adipocytes (data not shown). To assess whether the gene expression profiles of lipid-overloaded adipocytes were linked with cytokine secretion, enzyme-linked immunosorbent assays (ELISAs) and chemotaxis assays were performed. As expected, conditioned media from SFA-treated adipocytes contained higher levels of proinflammatory cytokines, including TNF-α, MCP-1, and IL-6, than did conditioned media from control adipocytes or MUFA-treated adipocytes (Fig. 2C). Next, to confirm the idea that SFA treatment could indeed potentiate inflammatory responses in hypertrophic adipocytes, chemotaxis assays were performed. The migration of THP-1 monocytes was increased when the cells were incubated with conditioned media from SFA-treated adipocytes (Fig. 2D). In contrast, conditioned media from MUFA-treated adipocytes had little effect on THP-1 migration (Fig. 2D). These data indicate that SFA-treated hypertrophic adipocytes tend to be proinflammatory, unlike MUFA-treated hypertrophic adipocytes.

FIG 2.

SFA-treated adipocytes, but not MUFA-treated adipocytes, promote proinflammatory responses. (A) After treatment of 3T3-L1 adipocytes with various long-chain fatty acids (500 μM), nuclear extracts and cell lysates were subjected to immunoblot analysis. CTL, control (BSA); PA, palmitic acid; SA, stearic acid; OA, oleic acid; TNF-α concentration, 10 ng/ml. (B) mRNA levels of proinflammatory genes were measured by quantitative RT-PCR. Relative mRNA levels were quantified after normalization against cyclophilin. *, P < 0.05 versus CTL cells (Student's t test). (C) Levels of TNF-α, MCP-1, and IL-6 secreted from lipid-overloaded adipocytes. (D) Migration of THP-1 monocytes. THP-1 monocytes were prestained with CellTracker (red) and incubated for 6 h in Transwell plates (8-μm pore) with conditioned medium (top left). Photomicrographs of migrated cells were taken (bottom left). Cell migration was assessed (right). After 3 or 6 days of fatty acid treatment, adipocytes were washed with PBS and incubated with fresh culture medium for 15 h. The conditioned medium was collected for ELISA and Transwell culture. *, P < 0.05 versus CTL cells (Student's t test).

Hypertrophic adipocytes are insulin resistant regardless of inflammatory status.

To assess insulin sensitivity of hypertrophic adipocyte models, we performed insulin-dependent glucose uptake assays. As shown in Fig. 3A, insulin-dependent glucose uptake was attenuated in hypertrophic adipocytes, regardless of the kinds of fatty acids used for treatment. Unexpectedly, in SFA- and MUFA-induced hypertrophic adipocytes, insulin downstream signaling pathways such as IRS-1 tyrosine phosphorylation and Akt phosphorylation were not different from those of control cells (Fig. 3B). In addition, when freshly differentiated 3T3-L1 adipocytes were challenged with conditioned media collected from SFA- or MUFA-induced hypertrophic adipocytes, it appeared that there was no significant reduction in insulin signaling cascades (Fig. 3C). Similarly, SFA- and MUFA-cotreated hypertrophic adipocytes were insulin resistant (Fig. 3D) without alteration of insulin downstream cascades (Fig. 3E), although they showed an intermediate level of proinflammatory gene expression compared with that of groups treated with SFA or MUFA alone (Fig. 3F). These data imply that insulin resistance of hypertrophic adipocytes might be dissociated from proinflammatory responses.

FIG 3.

Hypertrophic adipocytes are insulin resistant without any changes in the insulin downstream signaling cascade. 3T3-L1 adipocytes were differentiated and then cultured for another 6 days with various long-chain fatty acids (500 μM). CTL, control (BSA); PA, palmitic acid; SA, stearic acid; OA, oleic acid. (A) Insulin-dependent glucose uptake assays using [¹⁴C]deoxyglucose. *, P < 0.05 versus CTL cells (Student's t test). (B) Immunoblot analysis of adipocytes treated for 6 days with long-chain fatty acids, with or without insulin (10 nM). (C) Conditioned medium was collected from adipocytes overloaded with long-chain fatty acids for 6 days and then treated with newly differentiated 3T3-L1 adipocytes (top). Immunoblot analysis of adipocytes treated with conditioned medium (CM) for 48 h, with or without insulin (10 nM) (bottom). (D to F) Differentiated 3T3-L1 adipocytes were treated with palmitic acid (PA; 500 μM), palmitic acid and oleic acid mixture (PA+OA; 250 μM each), and oleic acid (OA; 500 μM). (D) Glucose uptake assays using [¹⁴C]deoxyglucose. ***, P < 0.001 versus CTL cells (Student's t test). (E) Immunoblot analysis of adipocytes treated for 6 days with long-chain fatty acids with or without insulin (10 nM) treatment. (F) The mRNA levels of proinflammatory genes were measured by qRT-PCR. Relative mRNA levels were quantified after normalization against cyclophilin. *, P < 0.05; **, P < 0.01; ***, P < 0.001 versus CTL cells (Student's t test).

To explore the potential relationship between adipocyte morphology and insulin resistance, we adopted single-cell-based insulin-dependent glucose bioprobe uptake assays (24). Consistent with the results from the radioisotope-based insulin-dependent glucose uptake assay (Fig. 3A), glucose bioprobe uptake analysis revealed that SFA- or MUFA-induced hypertrophic adipocytes were insulin resistant (Fig. 4A; see also Movies S1 and S2 in the supplemental material). Compared with control adipocytes, reduced uptake rates of glucose bioprobe were similar in SFA- or MUFA-challenged adipocytes (Fig. 4A). However, it is of interest that some small adipocytes among lipid-overloaded adipocytes showed relatively intense glucose bioprobe signals (Fig. 4A). To examine whether insulin-dependent glucose uptake was differentially regulated in response to adipocyte cell size and/or lipid droplet morphology in an inflammation-independent manner, we classified morphologically heterogeneous populations of MUFA-induced hypertrophic adipocytes into four categories: (i) small/multilocular adipocytes (S/M-ADs), (ii) large/multilocular adipocytes (L/M-ADs), (iii) small/uniloculus-like adipocytes (S/U-ADs), and (iv) large/uniloculus-like adipocytes (L/U-ADs) (Fig. 4B). As shown in Fig. 4C, the relative portion of L/U-ADs was predominantly increased by both SFA- and MUFA-overloaded hypertrophic adipocytes. Single-cell-based analyses with glucose bioprobe revealed that insulin-dependent glucose bioprobe uptakes in S/M-ADs and L/M-ADs were comparably increased after insulin stimulation. While the degree of insulin-dependent glucose bioprobe uptake in S/U-ADs was slightly lower than that of S/M-ADs or L/M-ADs, the level of insulin-dependent glucose bioprobe uptake in L/U-ADs was markedly attenuated (Fig. 4D). Together, these results suggest that adipocyte hypertrophy with uniloculus-like lipid droplets would lead to adipocyte insulin resistance in a cell-autonomous manner, independent of inflammatory responses.

FIG 4.

Hypertrophic adipocytes are insulin resistant. 3T3-L1 adipocytes were differentiated and then cultured for another 6 days with various long-chain fatty acids (500 μM). CTL, control (BSA); PA, palmitic acid; SA, stearic acid; OA, oleic acid. (A) Hypertrophic adipocytes challenged with fatty acids were incubated with a Cy3-labeled glucose bioprobe. After a 10-min incubation, insulin was added, and the fluorescence intensity of the glucose bioprobe was monitored every 2 min using a DeltaVision imaging system. Arrows designate relatively small adipocytes. Scale bars, 80 μm. ***, P < 0.001 versus CTL cells (ANOVA). (B) Hypertrophic adipocytes challenged with OA (500 μM; 6 days) were categorized into four groups: (i) S/M-ADs, <40 μm in diameter with multilocular lipid droplets; (ii) L/M-ADs, >40 μm in diameter with multilocular lipid droplets; (iii) S/U-ADs, <40 μm in diameter with uniloculus-like lipid droplets; and (iv) L/U-ADs, >40 μm in diameter with uniloculus-like lipid droplets. (C) Cellular distributions of each categorized cell type were computationally calculated by using ImageJ. (D) Glucose bioprobe fluorescence intensity in each cell type was detected every 2 min using a DeltaVision imaging system (bottom). For a single analysis, 50 images with ∼200 cells were analyzed. ***, P < 0.001 versus CTL cells (ANOVA).

Enlarged adipocytes with uniloculus-like lipid droplets showed defective GLUT4 trafficking.

Unlike SFA-induced hypertrophic adipocytes, MUFA-induced hypertrophic adipocytes did not alter the expression of proinflammatory genes (Fig. 2B). Despite this result, either of the SFA- or MUFA-induced hypertrophic adipocytes greatly diminished insulin-dependent glucose uptake ability (Fig. 3A and 4A). Thus, we asked the question whether GLUT4 trafficking might be associated with decreased insulin-dependent glucose uptake in hypertrophic adipocytes. Adipocytes expressing mCherry-GLUT4-Myc were immunostained with anti-Myc:GFP (where GFP is green fluorescent protein) antibodies, and GLUT4-Myc:GFP signals on the plasma membrane were detected using total internal reflection fluorescence microscopy (TIRFM), in the presence or absence of insulin (Fig. 5A). As shown in Fig. 5B, the amounts of plasma membrane GLUT4 in control adipocytes were increased upon insulin exposure. Interestingly, SFA- or MUFA-treated L/U-ADs decreased GLUT4 plasma membrane translocation in response to insulin exposure (Fig. 5B). In contrast, insulin-dependent GLUT4 translocation was intact in the other types of adipocytes (S/M-ADs, L/M-ADs, and S/U-ADs) (Fig. 5C).

FIG 5.

Insulin-stimulated GLUT4 trafficking is impaired in large/uniloculus-like adipocytes. (A to C) 3T3-L1 adipocytes stably expressing myc-GLUT4-mCherry were challenged with various fatty acids (500 μM) for 6 days. The cells were fixed, stained with an anti-Myc antibody (green), and imaged on a total internal reflection fluorescence microscopy (TIRFM) in the presence or absence of insulin (100 nM). CTL, control (BSA); PA, palmitic acid; SA, stearic acid; OA, oleic acid. Scale bars, 20 μm. (A) Schematic drawing illustrating events observed under the TIRF zone. (B) Insulin-induced GLUT4 membrane insertion was examined by using TIRFM in nonpermeabilized cells. Data presented are representative microscopic images of L/U-ADs in each indicated group. (C) 3T3-L1 adipocytes were treated with OA (500 μM) for 6 days. Insulin-induced GLUT4 membrane insertion was examined by using TIRFM in nonpermeabilized cells. Data presented are microscopic images representative of indicated groups. (D) Cellular actin structures were detected by phalloidin staining. After 6 days of OA (500 μM) treatment, 3T3-L1 adipocytes were fixed, permeabilized, and stained with phalloidin-TRITC (red), BODIPY (green), and DAPI (blue). Scale bars, 20 μm. (E) Stable 3T3-L1 adipocytes expressing GLUT4-mCherry were challenged with OA (500 μM) for 6 days. In the presence of insulin, adipocytes were fixed and imaged on a confocal microscope. Scale bars, 20 μm.

Since it is well known that cytoskeletal proteins are involved in adipogenesis, lipid droplet formation, and GLUT4 trafficking (29–31), we hypothesized that cytoskeleton development might be altered in hypertrophic adipocytes containing uniloculus-like lipid droplets, thus impeding GLUT4 docking to the plasma membrane. To test this hypothesis, lipid-overloaded hypertrophic adipocytes were stained with phalloidin-TRITC (where TRITC is tetramethyl rhodamine isocyanate) to detect the cellular actin organization. As shown in Fig. 5D, the cytosolic and cortical actin structures were markedly disorganized in L/U-ADs while those of S/M-ADs, L/M-ADs, and S/U-ADs were well organized. Consistent with the pattern of actin organization, insulin-dependent plasma membrane GLUT4 translocation was decreased in L/U-ADs compared with that in S/M-ADs, L/M-ADs, and S/U-ADs (Fig. 5E). These data suggest that the reduction of insulin-dependent glucose uptake in hypertrophic adipocytes would be the result, at least in part, of the dysregulation of cortical actin remodeling and the consequent impairment of insulin-dependent GLUT4 plasma membrane translocation.

TLR4 mutant mice fed short-term HFD exhibit adipose tissue insulin resistance without adipose tissue inflammation.

Proinflammatory responses and adipocyte hypertrophy are rapidly induced in early obesity in C57BL/6J mice fed a short-term (<1 week) HFD (14–18). To test whether adipocyte hypertrophy could promote insulin resistance in adipose tissue regardless of inflammatory responses, C3H/HeJ Toll-like receptor 4 (TLR4) mutant mice, defective in TLR4-dependent inflammation, were tested. After 1 week of HFD, body weight gain and the epididymal adipose tissue (eAT) mass were increased in both C3H/HeN wild-type control and C3H/HeJ TLR4 mutant mice (Fig. 6A). The adipocytes were also comparably enlarged in control and TLR4 mutant mice upon HFD (Fig. 6B). Unlike HFD-fed control mice, HFD-fed TLR4 mutant mice did not show elevated adipose tissue inflammation, despite adipose tissue expansion (Fig. 6C). To evaluate the systemic insulin sensitivity of control and TLR4 mutant mice, an oral glucose tolerance test was administered. As shown in Fig. 6D and E, both control and TLR4 mutant mice fed an HFD were glucose intolerant.

FIG 6.

TLR4 mutant mice fed a short-term HFD develop systemic insulin resistance. For 1 week, 10-week-old C3H/HeN and C3H/HeJ mice were fed an NCD or HFD. (A) Body weight (BW) gain, epididymal adipose tissue (eAT) mass, and liver mass were measured. (B) Distribution of adipocyte sizes in eAT. (C) mRNA levels of inflammatory genes from eAT were measured by quantitative real-time PCR analysis. Relative mRNA levels were quantified after normalization against cyclophilin. (D and E) Oral glucose tolerance test (D) and area under the curve (AUC) (E) results for all groups. Each bar represents the mean ± standard deviation (SD) for each group of mice (n = 7). *, P < 0.05; ***, P < 0.001; n.s., not significant.

To test whether hypertrophic adipocytes induced by short-term HFD feeding might be involved in glucose intolerance, we isolated adipose tissues from control and TLR4 mutant mice and performed ex vivo insulin-dependent glucose bioprobe uptake assays. As shown in Fig. 7A, the glucose bioprobe signals were detected in the cytosolic region of primary adipocytes upon administration of insulin. Despite decreased inflammatory responses in adipose tissue of TLR4 mutant mice, the degree of insulin-dependent glucose bioprobe uptake ability in TLR4 mutant mice upon HFD was similar to that of control mice (Fig. 7B and C). Taken together, these results suggest that hypertrophic adipocytes primarily cause adipose tissue insulin resistance in early obesity, independent of adipocyte inflammation.

FIG 7.

The adipose tissues of TLR4 mutant mice fed a short-term HFD are insulin resistant. Epididymal adipose tissues (eAT) from wild-type control C3H/HeN mice and TLR4 mutant C3H/HeJ mice fed an HFD for 1 week were ex vivo cultured. (A) The eATs of control C3H/HeN mice were incubated with CellTracker (top; red) or the glucose bioprobe (bottom; red) and BODIPY (blue) in the presence of insulin and visualized with confocal microscopy. Scale bars, 10 μm (left) and 5 μm (right; magnified images). (B and C) Insulin-dependent glucose bioprobe uptake assay. Adipose tissues were ex vivo cultured with or without insulin (1 μM). Glucose bioprobe (red) and BODIPY (blue) were incubated for 30 min. (B) The relative glucose bioprobe intensity per cell was analyzed using ImageJ. Each bar represents the mean ± SD for each group of mice (n = 7). *, P < 0.05; n.s., not significant. (C) Glucose bioprobe fluorescence from each group was visualized with confocal microscopy. Data presented are representative microscopic images. Scale bars, 200 μm.

Rosiglitazone generates small adipocytes and stimulates insulin-dependent glucose uptake.

As peroxisome proliferator-activated receptor γ (PPARγ) agonists, thiazolidinediones (TZDs) have multiple roles, such as insulin sensitization, anti-inflammatory effects, and induction of new adipogenesis in obese animals (32, 33). To determine whether newly differentiated small adipocytes might be more insulin sensitive than nearby hypertrophic adipocytes, rosiglitazone, one of the TZD class of drugs, was administered orally to obese db/db mice. As expected, rosiglitazone restored the mRNA level of adiponectin and insulin signaling accompanied with novel adipocyte differentiation in the adipose tissue of db/db mice (Fig. 8A and B). To discriminate dying adipocytes in crown-like structures (CLS) from newly differentiated adipocytes, adipose tissues were immunostained with CD11b antibody. As shown in Fig. 8C, newly differentiated small adipocytes did not show CD11b-positive signals but showed a morphology distinct from that of the dying adipocytes in CLS. In db/db mice, ex vivo assays showed that insulin-dependent glucose bioprobe uptake was greatly increased in rosiglitazone-induced, newly differentiated small adipocytes compared to the uptake in adjacent large adipocytes (Fig. 8D). These in vivo and ex vivo results indicate that hypertrophic adipocytes with large/unilocular lipid droplets in obese adipose tissue are important for the induction of adipose tissue insulin resistance in obesity.

FIG 8.

Rosiglitazone-induced, newly differentiated small adipocytes are insulin sensitive. For 1 month, 12-week-old db/+ and db/db mice were treated without or with rosiglitazone (15 mg/kg) by oral gavage. (A) Relative mRNA levels of adiponectin and TNF-α from the epididymal adipose tissues (eAT) of rosiglitazone-treated db/+ and db/db mice. *, P < 0.05; **, P < 0.01. (B) Immunoblot analysis of epididymal adipose tissue (eAT). Phosphorylation of Akt (Ser308) with or without insulin (50 nM) treatment. (C) Whole-mount immunohistochemistry analysis of eATs of rosiglitazone-treated db/+ and db/db mice. Newly differentiated small adipocytes (top) and crown-like structure (bottom). Adipose tissues were stained with CD11b antibody (red), BODIPY (green), and DAPI (blue) (n = 4). Scale bars, 50 μm. (D) Ex vivo glucose bioprobe uptake assay. eATs from rosiglitazone-treated db/+ or db/db mice were ex vivo cultured with or without insulin (1 μM). Glucose bioprobe (red) and BODIPY (green) were incubated for 30 min (n = 4). Data presented are representative microscopic images. Scale bars, 100 μm.

DISCUSSION

In obesity, increased adiposity and immune cell infiltration into adipose tissue contribute to insulin resistance in parallel with dysregulation of glucose and lipid metabolism (5, 6, 8, 15). Given that adipose tissue is composed of various cell types, including endothelial cells, macrophages, lymphocytes, and preadipocytes in stromal vascular factions, as well as adipocytes, it has been difficult to determine whether adipocyte hypertrophy per se might induce adipose tissue insulin resistance independent of proinflammatory responses in obesity. The recent findings that insulin resistance induced by short-term HFD could occur in the absence of proinflammatory responses (15, 16, 34) suggest that adipocyte dysfunction associated with hypertrophy would play a crucial role in the incidence of adipose tissue insulin resistance. In this regard, the characterization of hypertrophic adipocytes is important to decipher the mechanism(s) of insulin resistance, especially in early obesity. Here, we demonstrated that adipocyte hypertrophy could directly cause insulin resistance in a cell-autonomous manner via dysregulation of cortical actin structures and impairment of GLUT4 trafficking.

Although there are many pieces of circumstantial evidence (35–38), the molecular mechanisms of adipocyte hypertrophy-induced insulin resistance are largely unknown. One of the major impediments to investigate hypertrophic adipocytes is the lack of proper model systems in vitro and in vivo. It is certain that primary adipocytes would be the best model to study adipocyte hypertrophy. However, there are several technical obstacles to using primary adipocytes as a representative of hypertrophic adipocytes in a model system. First, large primary adipocytes are physically and chemically fragile. Because of this weakness, collagenase digestion and isolation steps might decrease significant portions of large adipocytes. Second, primary adipocytes cannot be free from contamination of stromal vascular cells. In adipose tissue, adipocytes and other cell types, including macrophages and T cells, are strongly and physically stuck together. In this regard, isolating “pure adipocytes” free from stromal cell contamination is not technically feasible (35, 38). Third, since primary adipocytes float under cell culture conditions, it is not technically practical to subject them to various experimental conditions. Thus, we have utilized 3T3-L1 adipocytes as an alternative approach to investigate adipocyte morphology and its functions. Although differentiated 3T3-L1 adipocytes exhibit differences from in vivo adipocytes, such as multilocular lipid droplets and lower responsiveness to lipopolysaccharide (LPS), it appears that the 3T3-L1 adipocyte has its own merits as a “pure” fat cell model without the influences of other cells found in adipose tissue.

To mimic insulin resistance in vitro, 3T3-L1 adipocytes have been challenged with glucose oxidase, chronic insulin, TNF-α, dexamethasone, or various free fatty acids (39–41). However, because these protocols have been used to study the relatively acute effects of the stimuli on adipocyte function, they have failed to reveal the roles of adipocyte hypertrophy and lipid droplet locularity. To overcome these limitations, we assessed the time- and dose-dependent effects of long-chain fatty acid treatment on 3T3-L1 adipocytes and optimized the conditions to generate hypertrophic adipocytes. We also found that long-chain fatty acid challenge induced hypertrophy of adipocytes differentiated from 3T3-F442A, 10T1/2, and stromal vascular cells (data not shown). In adipocytes, it has been reported that SFAs stimulate proinflammatory responses through TLRs with JNK and NF-κB activation and block the insulin signaling cascade (42–44). In accordance with previous data, SFA-challenged hypertrophic adipocytes increased proinflammatory responses (Fig. 2).

In this study, when adipocytes were chronically treated with free fatty acids for 6 days, the enlargement in fat cell size and the defect in insulin-dependent glucose uptake were similar in MUFA- and SFA-challenged hypertrophic adipocytes (Fig. 1, 3A, and 4). To rule out the potential involvement of SFA-induced inflammatory responses in the regulation of insulin sensitivity, MUFA-induced hypertrophic adipocytes were subjected to investigate the correlation between lipid droplet morphology and insulin sensitivity. We observed that insulin-dependent glucose uptake was less efficient in L/U-ADs than in S/M-ADs, S/U-ADs, and L/M-ADs. These results imply that impaired insulin sensitivity would be induced by excess lipid accumulation with enlarged lipid droplets and cellular hypertrophy rather than inflammation.

Cortical actin assembly is essential for the maintenance of proper cell shape, motility, and many other cellular functions (45, 46). In adipocytes, cortical actin remodeling is associated with adipogenesis, triglyceride accumulation, lipid droplet formation, and GLUT4 trafficking (29, 30). In particular, it has been shown that cortical actin assembly is involved in GLUT4 storage vesicle docking and tethering to the plasma membrane, while microtubule organization is associated with the approach of GLUT4 storage vesicles (31). GLUT4 is the major insulin-regulated glucose transporter; it coordinates insulin action in adipose tissue and muscle (47–49). Adipose tissue-selective ablation of GLUT4 impairs insulin sensitivity in metabolic tissues, and adipose-selective overexpression of GLUT4 enhances glucose disposal and increases fat cell number (hyperplasia) (50, 51). Microarray analysis of primary adipocytes from mice fed a short-term HFD showed that several cytoskeletal genes were upregulated or downregulated (microarray data are available under GEO accession number GSE65557). Thus, given that adipocyte morphology is dramatically changed by cell size and/or lipid droplet locularity, it is plausible to speculate that cortical actin structures and/or microtubules that mediate GLUT4 translocation might be involved in adipocyte hypertrophy-induced insulin resistance. Here, we demonstrated that the amounts of GLUT4 translocated to the plasma membrane upon insulin clearly decreased in L/U-ADs (Fig. 5C). In addition, cortical actin structures were disrupted in L/U-ADs, while microtubules were relatively intact (data not shown), indicating that decreased GLUT4 trafficking to the plasma membrane in L/U-ADs might be due to the disrupted cortical actin structure. However, we cannot exclude the possibility that the disruption in the membrane fusion of GLUT4 storage vesicles, which is modulated by SNAREs, also contributes to impaired GLUT4 trafficking and insulin resistance in L/U-ADs.

TLR4 is a pattern recognition receptor that plays a key role in the innate immune system by activating proinflammatory signaling pathways. C3H/HeJ mice, which have a missense mutation in the third exon of the TLR4 gene, have been widely used to investigate the role of TLR4 in the regulation of innate immunity (52, 53). Several TLR4-defective animal models exhibit different metabolic phenotypes in obesity, which result from different HFD regimes, the animals' sex, different background strains, and various TLR4 mutations such as null and missense (54–58). Nonetheless, most reports have consistently shown that C3H/HeJ mice are resistant to HFD-induced inflammation in adipose tissue. Here, to avoid complex circumstances in severe obesity and minimize the influences of immune cells, adipose tissues from short-term HFD-fed mice with adipocyte hypertrophy were assessed for ex vivo study. Although C3H/HeJ mice were protected from proinflammatory responses upon short-term HFD, the degree of adipocyte hypertrophy and insulin resistance was similar to that of the control C3H/HeN mice (Fig. 6 and 7). These data indicate that adipocyte enlargement might be a crucial factor for inflammation-independent insulin resistance in vivo.

Several lines of evidence support our idea that adipocyte hypertrophy is associated with insulin resistance in vivo, regardless of inflammation. First, in the case of Cushing's syndrome, patients with excess glucocorticoid develop central obesity and insulin resistance with suppressed immune responses. Second, the reduction of early B cell factor 1 (EBF1) increases adipocyte size and provokes insulin resistance but does not influence the inflammatory pathways (21). Third, our previous study shows that short-term HFD challenge in JNK1 knockout mice, Rag1 knockout mice (T-cell and B-cell depletion), and clodronate-treated mice (phagocytic macrophage depletion) induces adipocyte hypertrophy and insulin resistance independent of inflammatory responses (16).

In adipose tissue, adipocytes have parenchymal functions consisting of conducting energy storage and energy release according to nutritional status. Unlike other cell types, adipocytes are able to buffer a certain range of energy fluctuation for maintenance of whole-body energy homeostasis. However, when chronic energy surplus overcomes the buffering capacity of adipocytes, it appears that the adipocyte would lose its parenchymal functions, resulting in insulin resistance as well as lipid dysregulation. In agreement with this idea, we showed that the changes in adipocyte size and lipid droplet morphology, which may represent the flexibility of adipocyte capacity, are closely associated with insulin resistance, probably through cytoskeletal remodeling and GLUT4 trafficking. In adipocytes, it is likely that the nature of lipid metabolites would differentially affect the immune responses with different degrees of inflammatory responses, while adipocyte insulin resistance would not be affected by the nature of the stored lipid metabolites. We and others have proposed a provocative idea that the initial adipose tissue immune response may participate in adipose tissue remodeling rather than insulin resistance to accommodate the changes occurring in early obesity (16, 34, 59). In contrast, intense adipose tissue inflammation actively contributes to systemic insulin resistance in severe obesity, when the parenchymal functions of adipose tissue are rigorously impaired.

To date, most studies of adipose tissue insulin resistance have focused on investigating causal factors in prolonged or severe obesity. This approach may overlook the adipocyte-autonomous role in adipose tissue insulin resistance. For instance, it appears that adipose inflammation may reflect compensatory responses to a long-term imbalance in energy homeostasis. In this study, we demonstrated that adipocyte hypertrophy indeed induced insulin resistance, partly through impaired GLUT4 trafficking, concomitant with cortical actin disorganization (Fig. 9). In addition, we observed that adipocyte hypertrophy-induced insulin resistance appeared to be uncoupled from proinflammatory responses at the early stage of obesity. In conclusion, our data suggest that adipocyte hypertrophy per se would be a primary determinant of adipose tissue insulin resistance in early obesity, independent of inflammatory responses.

FIG 9.

Graphical representation of the process whereby long-chain fatty acid challenge induces adipocyte hypertrophy and lipid droplet unilocularization. Hypertrophied adipocytes with uniloculus-like lipid droplets show decreased GLUT4 translocation to the plasma membrane and disorganized cortical actin structures. Hypertrophy-mediated impairment of GLUT4 translocation results in adipocyte insulin resistance regardless of inflammation.