Macrophage VLDLR mediates obesity-induced insulin resistance with adipose tissue inflammation
Obesity is closely associated with increased adipose tissue macrophages (ATMs), which contribute to systemic insulin resistance and altered lipid metabolism by creating a pro-inflammatory environment. Very low-density lipoprotein receptor (VLDLR) is involved in lipoprotein uptake and storage. However, whether lipid uptake via VLDLR in macrophages affects obesity-induced inflammatory responses and insulin resistance is not well understood. Here we show that elevated VLDLR expression in ATMs promotes adipose tissue inflammation and glucose intolerance in obese mice. In macrophages, VLDL treatment upregulates intracellular levels of C16:0 ceramides in a VLDLR-dependent manner, which potentiates pro-inflammatory responses and promotes M1-like macrophage polarization. Adoptive transfer of VLDLR knockout bone marrow to wild-type mice relieves adipose tissue inflammation and improves insulin resistance in diet-induced obese mice. These findings suggest that increased VLDL-VLDLR signaling in ATMs aggravates adipose tissue inflammation and insulin resistance in obesity.
Obesity is characterized by chronic and low-grade inflammation accompanied with macrophage accumulation in adipose tissue, eventually leading to metabolic disorders including insulin resistance and type 2 diabetes1, 2. Adipose tissue macrophages (ATMs) are key players in adipose tissue inflammatory responses in obesity3,4,5. In lean animals, the large number of ATMs is composed of alternatively activated (M2-like) macrophages expressing high levels of interleukin (IL)-4, -10, -13, and arginase (ARG) 1 that are associated with insulin sensitivity6, 7. Although it has been shown that M2-like macrophages might produce catecholamines to enhance adaptive thermogenesis8,9,10, a very recent study reported that M2-like macrophages do not produce catecholamines11. These controversial findings are needed to be further investigated. In contrast, in obese animals, the population of classically activated (M1-like) macrophages is rapidly increased in adipose tissue12, 13. M1-like ATMs secrete numerous pro-inflammatory cytokines, such as tumor necrosis factor alpha (TNFα), and MCP-1, which aggravates adipose tissue inflammation and insulin resistance in obesity14, 15. In obese adipose tissue, pro-inflammatory cytokines secreted from M1-like ATMs induce adipokine dysregulation and impair insulin action to confer systemic insulin resistance16, 17. Thus, the imbalance between M1- and M2-like ATMs plays an important role to modulate pro-inflammatory responses in obese adipose tissue.
M1- or M2-like polarization of ATMs has been attributed to dynamic changes in adipose tissue microenvironments18. Concurrent with the expansion of adipose tissue in obesity, ATMs participate in adipose tissue remodeling by storing surplus lipid metabolites, giving rise to a subpopulation of lipid-laden ATMs19,20,21,22. Recent studies have shown that cytotoxic lipid species, such as free cholesterol, and short-chain saturated fatty acids, are elevated, whereas protective lipid metabolites, such as long-chain polyunsaturated fatty acids, are decreased in the lipid-laden ATMs of obese mice21, 23. Furthermore, lipid-overloaded macrophages in obese adipose tissue stimulate pro-inflammatory cytokines such as TNFα, steering to insulin resistance24. Also, it has been reported that macrophages are able to produce anti-inflammatory lipid metabolites such as DHA and EPA25. These findings suggest that alteration of lipid metabolism in ATMs would be crucial to induce inflammatory responses and insulin resistance in obesity.
In plasma, essential lipid metabolites, such as cholesterol and triglycerides, are circulated in the form of lipoproteins26. Major triglyceride-carrying lipoproteins are very low-density lipoprotein (VLDL) and chylomicron27. VLDL receptor (VLDLR) has a pivotal role to uptake VLDL and chylomicron through receptor-mediated endocytosis or lipoprotein lipase (LPL)-dependent lipolysis28, 29. VLDLR, a member of the low-density lipoprotein (LDL) receptor (LDLR) family, is abundantly expressed in adipose tissue, heart, kidneys, and skeletal muscle28, 29. Patients with VLDLR mutations exhibit low body mass index (BMI) compared to normal subjects30, 31. Similarly, VLDLR-deficient mice are protected from high fat diet (HFD)-induced obesity32. Furthermore, VLDLR-deficient mice exhibit improved glucose intolerance upon HFD, accompanied with alleviated inflammation and ER stress in adipose tissue33. In addition, it has been suggested that VLDL might influence cellular inflammatory responses in macrophages, thereby potentiating metabolic complications, such as atherosclerosis and diabetes34, 35. However, it remains largely unknown whether VLDLR-mediated VLDL uptake in macrophages is an important factor in mediating adipose tissue inflammation and insulin resistance in obesity.
In this study, we demonstrate that VLDLR is elevated in obese ATMs, and promotes adipose tissue inflammation by upregulating ceramide production and facilitating M1-like macrophage polarization. Moreover, bone marrow transplantation (BMT) from VLDLR knockout (KO) mice into wild-type (WT) recipient mice attenuates insulin resistance in diet-induced obesity (DIO), simultaneously with reduced adipose tissue inflammation. Altogether, our data suggest that upregulated macrophage VLDLR could provoke insulin resistance by enhancing pro-inflammatory signaling pathways, accompanied with altered lipid profiles under lipid-rich conditions in obesity.
VLDLR is abundantly expressed in adipose tissue28, 29. However, it is largely unknown whether VLDLR expression might be altered in obese adipose tissue. To address this, VLDLR expression was examined in EATs from lean and obese mice. Compared to lean EATs, the level of VLDLR mRNA was elevated in obese EATs (Fig. 1a, b). mRNA levels of other LDLR family members, including LDLR and apolipoprotein receptor (ApoER) 2, were not significantly changed in EATs of obese mice (Fig. 1a, b). As positive controls, the mRNA levels of TNFα and Acrp30 were examined. In accordance with a previous report36, the level of VLDLR mRNA in human adipose tissue showed a positive correlation with BMI (Supplementary Fig. 1). To further characterize the expression patterns of adipose tissue VLDLR, EATs were fractionated into adipocytes and SVCs. Unlike in adipocytes, the level of VLDLR mRNA was elevated in SVCs from HFD-induced obese mice as compared to those from control animals (Fig. 1c). To verify high expression of VLDLR in SVCs of DIO, SVCs were further separated into F4/80 and CD11b double-positive ATMs by using fluorescence-activated cells sorting. Upon HFD, the level of VLDLR mRNA was clearly raised in F4/80+ and CD11b+ ATMs, while those of LDLR and ApoER 2 were not altered (Fig. 1d). Moreover, elevated VLDLR protein was detected in CD11b+ ATMs from obese adipose tissues (Fig. 1e). In DIO, the levels of VLDLR mRNA were also elevated in peritoneal and liver macrophages (Supplementary Fig. 2a, b). On the other hand, the level of VLDLR mRNA in liver macrophages was quite low (Supplementary Fig. 2b). Together, these results indicated that VLDLR is highly expressed in obese adipose tissue, particularly in ATMs.
To investigate whether macrophage VLDLR might contribute to storage of intracellular lipid metabolites, VLDLR was overexpressed in peritoneal macrophages (Fig. 2a, b). As shown in Fig. 2c, d, the level of intracellular triglycerides was increased in VLDLR-overexpressing macrophages in the presence of VLDL, while that of cholesterol was not altered. Given the high expression of VLDLR in obese ATMs, we next tested whether VLDLR in ATMs might be involved in pro-inflammatory responses. In VLDLR-overexpressing macrophages, the presence of VLDL stimulated the expression of pro-inflammatory marker genes, including iNOS, TNFα, monocyte chemoattractant protein (MCP)-1, serum amyloid A (SAA), IL-1β, and interferon (IFN)γ (Fig. 2e). These results suggested that elevation of macrophage VLDLR expression would stimulate pro-inflammatory responses in the presence of VLDL, simultaneously with intracellular triglycerides accumulation.
As VLDLR-overexpressing macrophages had elevated intracellular triglycerides in the presence of VLDL (Fig. 2), we investigated whether macrophages could uptake VLDL via VLDLR. To address this, peritoneal macrophages isolated from WT or VLDLR KO mice were challenged with fluorescence-conjugated VLDL (VLDL-DiI). As illustrated in Fig. 3a, b, VLDLR KO macrophages hardly took up VLDL compared to WT macrophages. While WT macrophages accumulated intracellular triglycerides with VLDL in a time-dependent manner, VLDLR KO macrophages marginally increased intracellular triglycerides (Fig. 3c). Intracellular cholesterol did not differ between WT and VLDLR KO macrophages with or without VLDL (Fig. 3d). It has been reported that VLDL would be uptaken by receptor-mediated endocytosis or lipoprotein lipase (LPL)-dependent lipolysis28, 29. To test whether LPL might be involved in VLDL uptaking, we investigated LPL expression and its enzymatic activity in WT and VLDLR KO macrophages. As shown in Supplementary Fig. 3, the levels of LPL mRNA and its enzymatic activity in macrophages were not associated with VLDLR expression. Moreover, suppression of LPL via siRNA did not significantly alter cellular triglycerides and cholesterol contents in the absence or presence of VLDL. To validate the potential roles of macrophage VLDLR, peritoneal macrophages from WT or VLDLR KO mice were treated with or without VLDL and subjected to gene expression profiling. In the presence of VLDL, VLDLR KO macrophages did not have augmented expression of various pro-inflammatory marker genes, such as iNOS, TNFα, IL-6, IL-1β, MCP-1, and SAA, while WT macrophages did show stimulated expression of these pro-inflammatory genes (Fig. 3e). These results imply that macrophages would uptake VLDL via VLDLR and potentiate inflammatory responses, concomitantly with intracellular triglycerides accumulation. It has been well established that cytokines produced from macrophages could impair insulin action in adipocytes16, 17. Thus, we speculated that deficiency of macrophage VLDLR might affect insulin-induced glucose uptake and insulin signaling in adipocytes. To address this, conditioned media (CM) were collected from VLDL-treated peritoneal macrophages isolated from WT or VLDLR KO mice and were treated to differentiated 3T3-L1 adipocytes. Compared with CM from VLDL-treated WT macrophages, CM from VLDL-treated VLDLR KO macrophages slightly but substantially enhanced insulin-stimulated glucose uptake ability and increased the level of glucose transporter 4 (GLUT4) mRNA (Fig. 3f and Supplementary Fig. 4). Furthermore, the phosphorylation levels of AKT and GSK3β were elevated in adipocytes treated with CM from VLDLR KO macrophages upon insulin (Fig. 3g). Together, these results indicated that macrophage VLDLR could mediate pro-inflammatory responses in the presence of VLDL, which would aggravate insulin action in adipocytes.
The findings that macrophage VLDLR expression was upregulated in obese adipose tissue and stimulated pro-inflammatory gene expression upon VLDL promoted us to test whether VLDLR might be abundantly expressed in either M1- or M2-like ATMs. ATMs were fractionated into M1-like ATMs (F4/80+, CD11b+, and CD11c+) and M2-like ATMs (F4/80+, CD11b+, and CD11c−). Compared with M2-like ATMs, M1-like ATMs more abundantly expressed VLDLR mRNA as well as TNFα and CD11c mRNAs (Fig. 4a). To gain further insights in the role of macrophage VLDLR, LPS, and IFNγ or IL-4, were added to cultured BMDMs to induce M1- or M2-like macrophage polarization, respectively. In BMDMs, VLDLR mRNA and protein were upregulated under M1-like macrophage-driving condition rather than under M2-like macrophage-driving condition (Fig. 4b, c). As positive controls, TNFα and arginase 1 (ARG1) expression were measured for M1-driving and M2-driving conditions, respectively (Fig. 4b, c). As VLDL was uptaken through macrophage VLDLR (Fig. 3), VLDL-DiI was incubated in BMDMs during M1- or M2-like macrophage polarization. As shown in Fig. 4d, M1-derived BMDMs accumulated more VLDL-DiI than did M2-derived BMDMs, implying that M1-like macrophages would uptake and store more VLDL due to elevated VLDLR expression. Next, to validate whether VLDLR might contribute to promote M1-like macrophage polarization in the presence of VLDL, BMDMs from WT or VLDLR KO mice were induced to M1- or M2-like macrophage phenotype with or without VLDL incubation. In the presence of VLDL, M1-derived BMDMs from WT mice further increased the expression of M1 marker genes, such as iNOS, TNFα, and IL-1β (Fig. 4e). On the contrary, in M2-derived BMDMs from either WT or VLDLR KO mice, the mRNA levels of M1 and M2 marker genes were not altered with or without VLDL (Fig. 4f and Supplementary Fig. 5a). These data suggested that macrophage VLDLR could potentiate M1-like macrophage polarization by uptaking VLDL.