TREM2 regulates obesity-induced insulin resistance via adipose tissue remodeling in mice of high-fat feeding

Background Adipose tissue remodeling plays a significant role in obesity-induced insulin resistance. Published studies reported that level of trigger receptor expressed on myeloid cells 2 (TREM2) in adipose tissue is up-regulated in animal models of obesity. This study aims to investigate whether TREM2 regulates obesity-induced insulin resistance via modulating adipose tissue remodeling in mice of high-fat diet (HFD). Methods Wild-type (WT) and TREM2−/− mice were both fed with a controlled-fat diet (CFD) or HFD for 12 weeks and studied for obesity and insulin resistance. Meanwhile, epididymal adipose tissue (EAT) was examined for morphological and pathological changes to determine adipose tissue remodeling. After that, adipocyte-derived MCP-1 was measured in adipocytes, adipose tissue and circulation. Next, inflammatory cytokines were determined in adipose tissue macrophages (ATM). At last, livers were analyzed for hepatic steatosis. Results TREM2−/− mice on HFD had increased obesity and insulin resistance compared with WT counterparts. Adipose tissue from TREM2−/− mice exhibited reduced mass but greater adipocyte hypertrophy and increased adipocyte death. Besides, adipocyte-derived MCP-1 was down-regulated in TREM2−/− mice, and circulating MCP-1 level was lower than that of WT mice. Furthermore, TREM2−/− mice displayed reduced infiltration of F4/80+CD11c+ macrophages into adipose tissue, which was unable to form crown-like structures (CLS) to clean dead adipocytes and cellular contents. Also, TREM2 deficiency augmented inflammatory response of adipose tissue macrophages in HFD mice. In addition, TREM2−/− mice demonstrated more severe hepatic steatosis than WT counterparts under HFD feeding. Conclusions Trigger receptor expressed on myeloid cells 2 may function as a feedback mechanism to curb obesity-induced insulin resistance via regulating adipose tissue remodeling.


Background
Obesity is highly prevalent all over the world, due to changes in our lifestyle and diet [1,2]. It is estimated that obesity has affected more than 600 million adults and 100 million children at present [2]. Accumulating evidence has demonstrated that obesity is the most important and common cause of insulin resistance [3,4]. During obesity, adipose tissue changes the number and size of adipocytes. In the meantime, cells of various types in stromal vascular fraction (SVF) of adipose tissue, especially adipose tissue macrophages (ATM), undergo numerical and functional changes. These biological processes are termed as "adipose tissue remodeling" [5]. Adipose tissue remodeling regulates physiological functions of adipose tissue, which plays a significant role during the pathogenesis and etiology of obesity-induced insulin resistance [5].
Trigger receptor expressed on myeloid cells 2 (TREM2), which belongs to the immunoglobulin superfamily of receptors [6], is mainly expressed on myeloid cells, such as macrophages [7], dendritic cells [8] and microglia [9,10] for regulating various cell biological behaviors including survival, proliferation, differentiation, phagocytosis and inflammatory response [6][7][8][9][10][11][12][13]. Recent studies have shown that TREM2 is also expressed on mature adipocytes [14]. Furthermore, TREM2 can act as a lipid sensing receptor to recognize and bind lipids [9]. In animal models of obesity, TREM2 gene expression was up-regulated in adipose tissue [14][15][16]. However, it is unknown whether TREM2 regulates obesity-induced insulin resistance via adipose tissue remodeling. In the present study, we tried to determine the effect of TREM2 gene deficiency on adipose tissue remodeling in mice of high-fat diet (HFD), and explored the effect of TREM2 on obesity-induced insulin resistance. We first examined obesity, insulin resistance and adipose tissue remodeling in TREM2 knockout (TREM2 −/− ) and wild-type (WT) mice under HFD challenge. Then, we determined adipocyte hypertrophy and adipocyte death of epididymal adipose tissue (EAT). Next, we explored numerical changes of macrophages and its underlying mechanism. After that, we measured inflammatory response of adipose tissue macrophages in HFD mice. Finally, we evaluated hepatic steatosis in mice under HFD feeding.

Animals and diets
All animal experiments in this study were approved by the Animal Care and Use Committee of Zhejiang University. WT mice of C57BL/6 were purchased from Shanghai SLAC Laboratory Animal Co. TREM2 −/− mice with the background of C57BL/6 were generously provided by Professor Macro Colonna (Department of Pathology and Immunology, School of Medicine) from Washington University in St. Louis [12]. All animals in this research were kept in the Laboratory Animal Center of Zhejiang University under an environmentally controlled condition, with temperature stable at 22 ± 2 °C, humidity stable at 55 ± 5% and a 12/12 h light/dark cycle. Male WT and TREM2 −/− mice with the age of 6 weeks and bodyweight of 21.0-23.0 g were both fed with HFD (D12492, 60% kcal of energy from fat, Research Diets) ad libitum for 12 weeks, and control WT and TREM2 −/− mice were fed with controlled-fat diet (CFD) (D12450B, 10% kcal of energy from fat, Research Diets). Food consumption was recorded twice a week and bodyweight was monitored weekly.

Insulin tolerance test (ITT) and glucose tolerance test (GTT)
For ITT, mice were fasted for 6 h before an intraperitoneal injection of insulin with the dosage of 0.8 U/kg bodyweight (for CFD mice) or 1.0 U/kg bodyweight (for HFD mice). For GTT, mice were fasted for 16 h before an intraperitoneal injection of glucose with the dosage of 1.5 g/kg bodyweight. Tail vein blood was collected at 0, 15, 30, 60, 90 and 120 min after injection. Glucose level was measured with a glucometer (Accu-Chek Aviva, Roche Diagnostics).

Tissue and blood collection
Mice were fasted overnight for 16 h before anesthetizing with 4% chloral hydrate. Blood was collected from retroorbital venous and spun down at 2500 g for 5 min under 4 °C. After separation, serum was stored under − 80 °C for further analysis. Epididymal fat pads were dissected and weighed. After rinsing in PBS, EAT was divided into three parts: one part was kept in a − 80 °C freezer; one part was fixed in 10% neutral-buffered formalin; while the last part was for adipocyte and macrophage isolation.

Insulin stimulation
To examine insulin signaling pathway activity in adipose tissue, mice were fasted for 6 h before anesthetizing with 4% chloral hydrate. Left EAT (marked as control) was removed, rinsed in PBS and flash frozen in liquid nitrogen. After that, abdominal cavity was closed up temporarily. Insulin was injected intraperitoneally with the dosage of 1.0 U/kg bodyweight for stimulation. Right EAT (marked as insulin-15 min) was harvested 15 min post injection and rinsed in PBS to remove possible contamination of insulin and flash frozen in liquid nitrogen. All EAT samples were transferred to a − 80 °C freezer for Western blotting.

Adipocyte and macrophage isolation and purification
Adipocyte and macrophage isolation protocol was based on previous publication with a minor modification [17]. Briefly, epididymal fat pads were minced into small pieces before incubating with collagenase type II (17101015, Gibico) on a 37 °C heated shaker for 40 min. Then, cell suspension was passed through a 100 micron filter. After repeated centrifugations of 1000g, supernatant layer (floating adipocytes) was collected from top. In the meantime, cell pellet (SVF) was resuspended in ACK Lysing Buffer to remove erythrocytes for further purification of macrophages.

RT-PCR
Total RNA extraction and RT-PCR were performed as published protocol [18]. The relative gene expression level was normalized to β-actin mRNA expression. Specific primers used in this research for β-actin, TNF-α, IL-1β, IL-6 and iNOS were listed in Table 1.

Western blotting
Samples of adipose tissue for detecting activity of insulin signaling pathway and samples of adipocytes for detecting MCP-1 expression were homogenized in cell lysis buffer supplemented with phenylmethylsulfonyl fluoride (PMSF) and protease inhibitor cocktail. After 10 min of incubation on ice, samples were centrifuged at 14,000 rpm under 4 °C for 15 min. Protein layer was extracted by a 1 ml syringe penetrating through floating lipid layer from the top. Protein content was determined by using bicinchoninic acid (BCA) assay. Samples contained 30 μg of protein were separated on 12% Bis-Tris gels (NP0343BOX, Thermo Fisher Scientific). Protein was wet transferred onto polyvinylidene difluoride (PVDF) membranes. Membranes were incubated with specific antibodies against AKT (9272, Cell Signaling Technology), p-AKT (9271, Cell Signaling Technology), MCP-1 (sc-52701, Santa Cruz Biotechnology) and β-tubulin (70-ab009-040, MultiSciences) under 4 °C overnight. All antibodies were diluted 1:1000 in 5% BSA. Membranes were then incubated with corresponding secondary antibodies for 1 h at room temperature. Western blots were developed by enhanced chemiluminescence (20-500-500, Biological Industries) and detected by X-ray films.

Histology and immunohistochemistry analysis
After fixing in 10% neutral-buffered formalin for 72 h, samples were dehydrated and paraffin embedded. EAT was sectioned into 4 μm sections and stained with hematoxylin and eosin (H&E) for morphological evaluation. The cross sectional area of each adipocyte was quantified by using ImageJ software. To determine live adipocytes, EAT sections were first stained with an antibody to perilipin (20R-PP004, Fitzgerald, 1:200 dilution with PBS) followed by a rabbit anti-guinea pig secondary antibody (ab6771, Abcam, 1:2000 dilution with PBS). Dead adipocytes, defined as adipocytes without positive perilipin expression as published before [19,20], were counted under random 200× microscopic fields and expressed as the percentage of total adipocytes of each image. To determine macrophage infiltration, EAT sections were first stained with an antibody to F4/80 (MCA497, AbD Serotec, 1:50 dilution with PBS) followed by a goat antirat secondary antibody (PV-9004, ZSGB-BIO).

Statistical analysis
All analyses were calculated with GraphPad Prism6 software. Results were expressed as the mean ± SEM. The statistical significance was identified with one-way ANOVA (Tukey test for post-hoc comparison) and Student's t test. A value of p < 0.05 was considered statistically significant.

TREM2 −/− mice exhibit increased obesity, promoted insulin resistance and altered adipose tissue remodeling under HFD feeding
To clarify the role of TREM2 deficiency on obesityinduced insulin resistance, we fed TREM2 −/− mice and their WT counterparts with CFD or HFD for 12 weeks. There was no significant difference in bodyweight between WT and TREM2 −/− mice under CFD. However, we observed a higher bodyweight in TREM2 −/− mice after HFD feeding compared with WT counterparts (39.4 ± 0.6 g in WT mice vs. 42.6 ± 0.7 g in TREM2 −/− mice at 8 weeks, N = 16/group, p < 0.001; 47.0 ± 0.4 g in WT mice vs. 49.4 ± 0.5 g in TREM2 −/− mice at 12 weeks, N = 16/group, p < 0.01) (Fig. 1a). To confirm whether the rise in bodyweight was due to an increased food intake, we compared food consumption of each group. All groups had similar levels of food consumption (p > 0.05), indicating the difference in bodyweight was independent of the amount of food intake (Additional file 1: Figure S1). To determine the effect of TREM2 deficiency on insulin resistance, we carried out GTT and ITT in both TREM2 −/− and WT mice on CFD and HFD. In CFD mice, TREM2 deficiency didn't alter glucose levels compared with WT in both GTT and ITT (Additional file 2: Figure S2). When HFD mice were challenged with insulin but not glucose, TREM2 −/− mice demonstrated higher blood glucose levels at 60, 90 and 120 min (p < 0.01) (Fig. 1b-e) compared with WT mice. Next, we analyzed P-Akt protein levels in EAT of WT and TREM2 −/− mice on HFD to determine the activity of insulin signaling pathway. TREM2 −/− mice showed lower level of Akt phosphorylation in Ser473 residue, as compared with WT mice, indicating a suppressed insulin signaling (Fig. 1f ), which was in accordance with worse ITT results. Consistently, TREM2 deficiency elevated fasting blood glucose levels of HFD mice after 12 weeks of feeding (14.33 ± 0.90 mmol/l in WT mice vs. 17.49 ± 1.19 mmol/l in TREM2 −/− mice, N = 13/group, p < 0.05) (Additional file 3: Figure S3D). The mass and percentage of bodyweight of epididymal fat pads were comparable between WT and TREM2 −/− mice fed with CFD, but were both reduced in HFD-fed TREM2 −/− mice compared with WT counterparts (2.38 ± 0.09 g in WT mice vs. 1.99 ± 0.09 g in TREM2 −/− mice, N = 10/ group, p < 0.01) (Fig. 1g) (5.23 ± 0.23% in WT mice vs. 4.15 ± 0.20% in TREM2 −/− mice, N = 10/group, p < 0.01) (Fig. 1h). The histopathology of EAT showed no difference between WT and TREM2 −/− mice on CFD. H&E staining revealed that adipocytes from TREM2 −/− mice under HFD feeding were considerably enlarged, while the number of macrophages was significantly reduced with fewer crown-like structures (CLS) (Fig. 1i).   (Fig. 2b), indicating TREM2 −/− mice have developed greater adipocyte hypertrophy. Adipocyte hypertrophy has been reported to drive adipocyte death, resulting in adipose tissue dysfunction and inflammation in the end [21]. To determine whether TREM2 deficiency will promote adipocyte death in HFD mice, EAT was immunohistochemically stained for perilipin, an adipocyte-specific cytomembrane protein. Immunohistochemistry revealed that there were fewer perilipin positive adipocytes in EAT of TREM2 −/− mice compared with WT counterparts, indicating a promoted adipocyte death (66.22 ± 0.90% of WT mice vs. 79.97 ± 2.16% of TREM2 −/− mice, N = 6/group, p < 0.001) (Fig. 2c, d). In addition, we observed less macrophage infiltration surrounding dead adipocytes in TREM2 −/− mice, so as CLS formation. Taken together, these results demonstrated that TREM2 deficiency promoted epididymal adipocyte hypertrophy and adipocyte death without stimulating macrophage infiltration in HFD mice.

TREM2 deficiency down-regulated MCP-1 expression of adipocytes in mice of HFD feeding
In obese state, monocytes were recruited from circulation and differentiated into F4/80 + CD11c + macrophages. Among all kinds of circulating chemokines released by adipose tissue, MCP-1 plays a significant role [23]. To explore whether suppressed F4/80 + CD11c + macrophage infiltration was due to lower concentration of MCP-1 in blood stream, circulating MCP-1 levels were detected via ELISA. Compared with WT mice, circulating MCP-1 level was significantly lower in TREM2 −/− mice on HFD (176.8 ± 13.9 pg/ml of WT mice vs. 126.1 ± 7.0 pg/ml of TREM2 −/− mice, N = 19/group, p < 0.01) (Fig. 4a). After that, MCP-1 expression was measured with RT-PCR in EAT of WT and TREM2 −/− mice. From RT-PCR analysis, we observed a reduced MCP-1 expression in TREM2 −/− mice (Fig. 4b). In adipose tissue, both macrophages and adipocytes can generate and secrete MCP-1. Therefore, MCP-1 expression was examined in both macrophages and adipocytes. It turned out that macrophages derived MCP-1 was comparable in both groups (Fig. 4c), while adipocytes MCP-1 expression was down-regulated in TREM2 −/− mice at both mRNA level (Fig. 4d) and protein level (Additional file 4: Figure S4). Taken together, these results demonstrated that TREM2 deficiency downregulated adipocyte-derived MCP-1 expression, leading to lower concentration of circulating MCP-1, which has a reduced effect to recruit monocytes infiltration.

TREM2 deficiency enhanced macrophage inflammatory response in mice of HFD feeding
Macrophages are the main source of pro-inflammatory cytokines in adipose tissue and play a pivotal role in the development of obesity-induced insulin resistance [24]. TREM2 has been known to exert anti-inflammatory effects via attenuating macrophage activation [12,13].
We proposed that TREM2 deficiency may up-regulate adipose tissue macrophage inflammatory response in mice of HFD feeding. To test this hypothesis, we isolated macrophages from epididymal fat pads of obese WT and TREM2 −/− mice to measure pro-inflammatory cytokines. Compared with WT counterparts, adipose tissue macrophages from TREM2 −/− mice underwent HFD expressed similar level of TNF-α (Fig. 5a), but higher levels of IL-1β (Fig. 5b), IL-6 ( Fig. 5c), and iNOS (Fig. 5d), indicating an enhanced inflammatory response.
Adipocyte hypertrophy is an important stress factor leading to adipocyte death; besides, hypertrophic adipocytes show features of necrosis as membrane rupture and functional membrane protein loss [19,20]. In our study, TREM2 −/− mice displayed higher incidence of adipocyte death (79.97 ± 2.16% of TREM2 −/− mice vs. 66.22 ± 0.90% of WT mice) (Fig. 2c, d), which could be a consequence of greater adipocyte hypertrophy.
Macrophages are the main source of pro-inflammatory cytokines in adipose tissue and play a pivotal role in the development of obesity-induced insulin resistance [24]. TREM2 has been known as an antiinflammatory regulator in immune process, since it can suppress inflammatory response via blocking Tolllike receptor signaling pathway [12,13]. In our study, we observed that macrophages of EAT expressed more pro-inflammatory cytokines such as IL-1β, IL-6 and iNOS in TREM2 knockout mice (Fig. 5).
Published work demonstrated that, down-regulation of TREM2 in adipose tissue in morbid obese patients is associated with advanced insulin resistance [27], which was in consistent with our experiment (Fig. 1d-f and Additional file 3: Figure S3D). Besides, elevated TREM2 expression was observed in obese animal models [14][15][16]. Hence, we hypothesize that TREM2 may act as a feedback protective mechanism to curb obesity inducedinsulin resistance via regulating adipose tissue remodeling. First, TREM2 alleviates adipocyte hypertrophy and adipocyte death via promoting adipogenesis. Next, TREM2 up-regulates adipocyte-derived MCP-1 expression to recruit F4/80 + CD11c + macrophage infiltration to isolate and clear dead adipocytes and cellular contents. In addition, TREM2 attenuates inflammatory response of macrophages in EAT under HFD feeding.
The present study has one major limitation that should be addressed. Our TREM2 −/− mice with the background of C57BL/6 were created according to traditional gene knockout technology [12]. In short, a portion of the transmembrane and cytoplasmic domains encoded by exons 3 and 4 was deleted in embryonic stem cells [12]. Because all cells in TREM2 −/− mice were TREM2 deficient, we can not distinguish whether TREM2 expressed on ATM or adipocytes plays a more important role in the pathogenesis and etiology of obesity-induced insulin resistance. Besides, traditional gene knockout technology allows for the possibilities that TREM2 expression on other tissue cells (yet to be discovered) may influence experimental results. Thus, an animal model with TREM2 conditional knockout (cellspecific knockout) in adipocytes and/or macrophages is warranted in future experiments to delineate the effect of TREM2 on obesity induced insulin resistance.

Conclusion
In conclusion, our data demonstrated that TREM2 may function as a feedback mechanism to inhibit obesityinduced insulin resistance. Our study suggested that TREM2 may act as a novel biomarker and potential therapeutic target of obesity and insulin resistance.