Inflammatory Pathways Regulated by Tumor Necrosis Receptor–Associated Factor 1 Protect From Metabolic Consequences in Diet-Induced ObesityNovelty and Significance
Rationale: The coincidence of inflammation and metabolic derangements in obese adipose tissue has sparked the concept of met-inflammation. Previous observations, however, suggest that inflammatory pathways may not ultimately cause dysmetabolism.
Objective: We have revisited the relationship between inflammation and metabolism by testing the role of TRAF (tumor necrosis receptor–associated factor)-1, an inhibitory adapter of inflammatory signaling of TNF (tumor necrosis factor)-α, IL (interleukin)-1β, and TLRs (toll-like receptors).
Methods and Results: Mice deficient for TRAF-1, which is expressed in obese adipocytes and adipose tissue lymphocytes, caused an expected hyperinflammatory phenotype in adipose tissue with enhanced adipokine and chemokine expression, increased leukocyte accumulation, and potentiated proinflammatory signaling in macrophages and adipocytes in a mouse model of diet-induced obesity. Unexpectedly, TRAF-1−/− mice were protected from metabolic derangements and adipocyte growth, failed to gain weight, and showed improved insulin resistance—an effect caused by increased lipid breakdown in adipocytes and UCP (uncoupling protein)-1–enabled thermogenesis. TRAF-1–dependent catabolic and proinflammatory cues were synergistically driven by β3-adrenergic and inflammatory signaling and required the presence of both TRAF-1–deficient adipocytes and macrophages. In human obesity, TRAF-1–dependent genes were upregulated.
Conclusions: Enhancing TRAF-1–dependent inflammatory pathways in a gain-of-function approach protected from metabolic derangements in diet-induced obesity. These findings identify TRAF-1 as a regulator of dysmetabolism in mice and humans and question the pathogenic role of chronic inflammation in metabolism.
Obesity and its metabolic complications, type 2 diabetes mellitus, hyperlipidemia, and hepatic steatosis, fuel cardiovascular mortality and morbidity.1 Many clinical and experimental observations have established an association between visceral obesity and a chronic inflammatory response in obese adipose tissue that enhances the risk of atherosclerosis, myocardial infarction, and stroke.2 Inflammation of visceral adipose tissue (VAT) results in the secretion of proinflammatory cytokines and chemokines from tissue-resident cells,2,3 which promote the recruitment of immune cells or activate inflammatory signaling networks.4 Although the temporal and functional dynamics of VAT-resident leukocytes have been well defined,5 the inflammatory signaling pathways that cause adipose tissue inflammation are less clear. Particularly, it remains controversial how chronic inflammation affects metabolism. For instance, neutralizing proinflammatory pathways driven by TNF (tumor necrosis factor)-α, IL (interleukin)-1β, or IL-6 has been reported to aggravate metabolic complications or not to be effective in clinical trials.6–8 These results suggest that proinflammatory signaling may not be exclusively pathogenic in obesity. We therefore revisited the role of TRAFs (tumor necrosis factor receptor–associated factors) in VAT inflammation. TRAFs represent intracellular signaling adapters that can bundle signaling events from upstream proinflammatory signals including TNF-α, IL-1β, and TLRs (toll-like receptors).9 TRAFs showed controversial outcomes in acute and chronic cardiometabolic inflammation,10,11 raising the possibility that inflammation may improve metabolism.
In This Issue, see p 639
Meet the First Author, see p 640
All data and methods used in the analysis and materials used to conduct the research will be made available to any researcher for the purpose of reproducing the results or replicating the procedures. All data, methods, and materials are available on personal request at the University Heart Center of Freiburg (D.W.).
A Methods section is available in the Online Data Supplement.
Genetic Deficiency of TRAF-1 Protects From Diet-Induced Obesity
To interrogate a role of TRAFs in VAT inflammation, we screened for gene expression of the known TRAFs 1 to 7 in epididymal VAT (epiVAT) of male C57BL/6J mice after consumption of a high-fat diet (HFD) or a standard diet (chow) for 12 weeks. Traf1 expression rose ≈10-fold, whereas the expression of other TRAFs was not or only modestly regulated (Figure 1A and 1B). TRAF-1 protein and gene expression were detectable in adipose tissue lymphocytes and adipocytes from epiVAT in confocal microscopy and quantitative polymerase chain reaction (Online Figures I and II). We observed a positive correlation of the Traf1 gene with Ccl2 (C-C motif chemokine ligand 2; MCP-1/[monocyte chemotactic protein 1] CCL2) and Adgre1 (macrophage marker F4/80; Figure 1C and 1D), suggesting that TRAF-1 is associated with adipose tissue inflammation.
We next aimed to clarify whether TRAF-1 directly contributes to diet-induced obesity (DIO) in rodents. Therefore, 8-week-old male wild-type (WT) and TRAF-1−/− mice consumed an HFD for 12 weeks. TRAF-1–deficient mice were protected from absolute and relative weight gain (Figure 1E through 1G; Online Figure III)—an effect caused by smaller peripheral fat depositions, decreasing fat mass in magnetic resonance imaging and magnetic resonance imaging body composition analysis, decreased ectopic lipid accumulation, and lowered levels of the adipocyte-specific leptin, but not by a decrease of lean tissue or organ size (Figure 1H through 1K; Online Figures III and IV; Online Table I). This loss of fat tissue was not caused by a lower food intake (Online Figure V). In accord with improved DIO, TRAF-1−/− mice demonstrated improved insulin resistance in an in vivo insulin challenge (Figure 1L and 1M) and diminished glucose and insulin levels in fasting mice (Figure 1N and 1O; Online Figure VI). Accordingly, gene expression for GLUT (glucose transporter)-2 (Slc2a2) and IRS (insulin receptor)-2 (Irs2) increased in TRAF-1−/− epiVAT (Online Figure VI). In addition, TRAF-1 deficiency improved liver steatosis (Figure 1P; Online Figure IV). Although TRAF-1−/− mice on a standard chow diet for 12 weeks had a comparable weight to WT mice, we detected a higher food intake, lower physical activity, and aggravated glucose utilization in these mice, rendering the beneficial metabolic effects in the HFD group as diet specific (Online Figure VII). Collectively, our findings identify that TRAF-1 contributes to DIO and its metabolic consequences.
TRAF-1 Deficiency Induces a Hyperinflammatory Phenotype in VAT
To test whether the metabolically beneficial phenotype of TRAF-1−/− mice translates into an improved inflammatory response, we quantified the number of leukocytes in epiVAT of WT and TRAF-1−/− mice. Surprisingly, TRAF-1–deficient mice presented higher numbers of leukocytes (Figure 2A) accumulating in crown-like structures in epiVAT (Figure 2B and 2C), suggesting enhanced inflammation. Flow cytometry of digested epiVAT confirmed higher numbers of obesity-specific CD11c+ proinflammatory adipose tissue macrophages (ATMs; Figure 2D through 2F), which was not caused by larger reservoirs of myeloid cells in the circulation, spleen, or bone marrow (Online Figure VIII). We observed an enhanced phosphorylation of the inflammatory MAP (mitogen-activated protein) kinase JNK (C-Jun N-terminal kinase; Figure 2G; Online Figure IX), higher epiVAT gene expression for the adipokines TNF-α, IL-6, and MCP-1 (Figure 2H), and for chemokines, TNF superfamily members, and TLRs in adipocytes from TRAF-1−/− epiVAT (Figure 2I and 2J). FACS (fluorescence-activated cell sorting)-sorted macrophages from epiVAT showed increased gene expression for Nos1, the coding gene for the driver for in vitro proinflammatory M1 polarization, iNOS (inducible NO synthase), and Ccl2, Il6, and Il12a (Figure 2K). Liver inflammation was not enhanced (Online Figure X), indicating a VAT-specific effect.
To address whether this phenotype in TRAF-1−/− macrophages was caused indirectly by adipocytes, we generated bone marrow chimeras by transplanting WT/CD45.1 and TRAF-1−/−/CD45.2 bone marrow into lethally irradiated WT/CD45.1 mice (Figure 2L). This strategy allowed us to selectively study TRAF-1−/− ATMs in a microenvironment with TRAF-1–competent adipocytes. Flow cytometry of digested epiVAT after 12 weeks of HFD revealed a relative increase of TRAF-1−/−CD11c+ ATMs (Figure 2M through 2O) with a proinflammatory gene expression (Figure 2P). ATMs in TRAF-1−/− mice showed amplified inflammatory gene expression after 4 weeks of HFD even in the absence of obesity (Online Figure XI), indicating that the hyperinflammatory phenotype precedes metabolic changes. In genetically chimeric mice generated by bone marrow transplantations, only these mice with a deficiency of TRAF-1 in hematopoietic cells, but not in stromal/vascular cells (including adipocytes), showed the hyperinflammatory VAT phenotype (Online Figures XII and XIII). These findings establish that TRAF-1 deficiency boosts ATM inflammation.
TRAF-1−/− Deficiency Induces Lipolysis and Increases Energy Expenditure
To assess how TRAF-1 deficiency protects from obesity, we asked how TRAF-1 alters adipocyte remodeling during obesity. We observed that the average adipocyte diameter and size distribution in sections of epiVAT shifted toward smaller adipocytes in TRAF-1−/− mice (Figure 3A through 3C). Traf1 mRNA abundance rose in differentiated adipocyte-like 3T3L-1 cells (Figure 3D) and in adipocytes from obese epiVAT (Figure 3E). Although we observed no relevant modulation of genes involved in lipid synthesis in epiVAT (data not shown), we detected increased plasma levels of the triglyceride metabolites free fatty acids (FFAs; Online Figure XIVA and XIVB) and an increase of the FFAs/triglyceride ratio in epiVAT/lysates from TRAF-1−/− mice (Online Figure XIVC), indicative of a higher lipid breakdown in TRAF-1−/− adipocytes.
Lipid breakdown of triglycerides into FFAs is regulated by adrenergic signaling and an upregulation of the lipolysis key enzymes HSL (hormone-sensitive lipase) and ATGL (adipocyte triglyceride lipase).12 After an intraperitoneal injection of isoproterenol, a β3-adrenoceptor agonist, FFAs and glycerol were higher in TRAF-1−/− compared with WT mice (Figure 3G and 3H), indicating that TRAF-1−/− adipocytes are more prone to lipolysis. This was confirmed in an in vitro lipolysis assay of epiVAT (Online Figure XIVD). Obese TRAF-1−/− adipocytes from epiVAT showed higher gene expression of HSL (Lipe) and ATGL (Pnpla2; Figure 3G), as well as more Perilipin, which is required for FFA trafficking (Figure 3I).
An excess of FFAs induces expression of UCP-1 (uncoupling protein-1) in brown adipose tissue (BAT), which boosts energy expenditure.13 We detected higher UCP-1 mRNA (Ucp1) in induces BAT (Figure 3J), augmented energy expenditure, and physical activity in obese TRAF-1−/− mice, but not in lean or TRAF-1−/− mice consuming chow (Figure 3K and 3L; Online Figures VII and XV). Notably, lipolysis in BAT itself was not changed in TRAF-1−/− mice compared with WT mice (Online Figure XIVE).
Adrenergic signaling and an excess of FFAs initiate thermogenesis.14 Accordingly, injection of the β3-agonist noradrenalin induced a higher body temperature in TRAF-1−/− compared with WT mice (Figure 3N). During cold adaption, TRAF-1−/− mice had a higher body temperature and increased expression of thermogenic genes in induced BAT (Figure 3M; Online Figure XVI), although we did not find consistently increased expression of genes indicating browning of adipose tissue in inguinal subcutaneous adipose tissue of TRAF-1−/− mice (Online Figure XVII). These data establish an enhanced lipid breakdown and increased energy expenditure as cause for improved obesity in TRAF-1−/− mice.
Finally, we asked how increased inflammation in TRAF-1−/− VAT contributes to this effect: first, neither a selective deficiency of TRAF-1 in bone marrow–derived nor in stromal/vascular cells sufficed to increase lipolytic gene expression to the extend observed in a global deficiency (Online Figure XVIII). Second, only inflammatory priming with TNF-α increased gene expression of adrenergic receptor-β3 (Adrb3) selectively in TRAF-1−/− adipocytes (Online Figure XIXA). Third, both adrenergic and inflammatory signaling synergistically induced proinflammatory gene expression in epiVAT (Online Figure XIXB). Fourth, in in vitro cultivation of epiVAT with conditioned media from macrophages, only the combination of TRAF-1−/− macrophages and TRAF-1−/− epiVAT induced lipolysis to a relevant extend (Online Figure XIXC). These findings indicate that TRAF-1–dependent inflammation in macrophages and lipolytic pathways in adipocytes act synergistically in the context of β-adrenergic signaling and lipolysis.
TRAF-1–Associated Signaling Pathways Are Enriched in Human Obesity
To clarify whether the TRAF-1 pathway associates with human obesity, we tested the enrichment of genes associated with the TRAF pathway in a gene array of human subcutaneous adipose tissue RNA from 7 healthy and 16 obese donors15 by gene set enrichment analysis. We found that genes in the TRAF-1 to -7 and in the TRAF-1 pathway (Online Table II) were higher expressed in obese donors (Figure 4A and 4B; false discovery rate<0.2). Accordingly, TRAF-1 expression in obese subcutaneous adipose tissue tissue lysates (Figure 4C) and isolated subcutaneous adipocytes (Figure 4D) was higher in donors with a higher body mass index. We next asked whether TRAF-1 expression could serve as biomarker for human obesity. We screened mRNA from peripheral blood mononuclear cells of participants of the TRAFICS study (Tumor Necrosis Receptor–Associated Factors in Cardiovascular Disease), a collective of 304 individuals with a high prevalence of the metabolic syndrome (Online Table III). We found that TRAF-1 expression was highest in patients diagnosed with ≥4 clinical components of the metabolic syndrome (Figure 4E). TRAF-1 correlated with the body mass index (Figure 4F) and fasting plasma lipids (Figure 4G and 4H), suggesting an association with human obesity and dyslipidemia.
A hallmark of adipose tissue inflammation is the accumulation of leukocytes and the expression of inflammatory mediators that perpetuate the inflammatory response by direct effects on tissue-resident cells or by enhanced leukocyte recruitment.4 Adipose tissue inflammation is associated with adipose tissue remodeling and insulin resistance. Disruption of immune cell function can improve obesity.5 However, it is not clear, whether inflammation is detrimental under every circumstance. This is best illustrated by the classical proinflammatory cytokine TNF-α that directly counteracts insulin signaling in adipocytes. Its inhibition reduced DIO in rodents,16 but the genetic knockout of its receptors aggravated VAT inflammation,8 raising the possibility that inflammation may in part be beneficial in DIO. Here, we have chosen to target TRAF-1 that inhibits several proinflammatory signaling cascades.9 Expectedly, TRAF-1 deficiency aggravated inflammation of obese VAT,10 but unexpectedly it improved metabolism, protected from obesity, and decreased ectopic lipid depositions by an induction of lipolytic pathways. This coincidence of increased inflammation and beneficial adipocyte remodeling is not entirely unexpected: Low-grade inflammation is required to protect from adipocyte dysfunction.17 Biologically, inflammation-induced lipolysis is needed to release energy resources during stress and infection.18 In line, TNF-α directly stimulates lipolysis in adipocytes.19 Lipolysis itself may also trigger inflammatory pathways by enhancing myeloid cell infiltration.20 Yet, rampant VAT inflammation may surpass these protective effects of low-grade inflammation.21
TRAF-1–deficient macrophages showed increased expression of inflammatory genes that have been attributed to an in vitro proinflammatory M1-like polarization, such as Nos1, Il12a, and Il6. Our findings indicate that the enhanced inflammatory cytokine secretion in TRAF-1−/− macrophages was necessary to increase ADRB3 expression on TRAF-1−/− adipocytes, leading to an enhanced responsiveness to adrenergic signaling. As a result, we detected higher lipolytic gene expression, lipolysis, and a higher release of the triglyceride metabolites glycerol and FFAs. FFAs increase thermogenesis in BAT by a UCP-1–dependent pathway,13 even in the absence of enhanced lipolysis in BAT itself.22 Accordingly, we detected more Ucp1 transcripts in BAT and enhanced thermogenesis.
To confer a metabolic benefit, a deficiency of TRAF-1 was necessary in both macrophages and adipocytes, whereas a selective deficiency in hematopoietic cells, including macrophages, sufficed to increase epiVAT inflammation, but did not protect from obesity. To validate these findings and to test whether a cell-specific inhibition of TRAF-1 would remain effective, future studies will have to use cell-specific strategies, for example, by Cre/lox-mediated conditional knockouts. This is also important because TRAF-1–dependent mechanisms seem to be tissue-specific: TRAF-1 deficiency aggravated lipopolysaccharide-induced lung inflammation,10 whereas TRAF-1−/− mice on an obesity-prone Ldlr−/− background mice were protected from atherosclerosis.11 Because atherosclerosis is driven by obesity and hyperlipidemia, TRAF-1−/− mice may primarily be protected by the improved metabolism independent of enhanced inflammation, or by a protective phenotype of TRAF-1–deficient macrophages specifically in the atherosclerotic plaque.11 The effects observed in our study seem to depend on an obese phenotype because we did not observe hyperinflammation in VAT or an improved metabolism in TRAF-1−/− mice consuming a chow diet. Yet, an increased food intake, lowered physical activity, and aggravated glucose utilization in these mice point toward a (modest) general role for TRAF-1 in affecting basal energy metabolism. However, these effects do not explain the phenotype of obese TRAF-1−/− mice. A panel of 1440-SNPs (single-nucleotide polymorphisms), which revealed a C57Bl/6J identity of >99.5%, furthermore ruled out an incongruent genetic background as potential bias.
Collectively, we present the unexpected finding that a deficiency of the inhibitory adapter TRAF-1 induced hyperinflammation in obese VAT and attenuated metabolic derangements. The underlying catabolic pathway was synergistically driven by inflammatory and adrenergic signaling. Our findings identify a dissociation of metabolism and rampant inflammation and propose that proinflammatory pathways in macrophages and lipolytic pathways in adipocytes are both limited by TRAF-1 under steady-state conditions. TRAF-1–dependent pathways were upregulated in human obesity, rendering our results relevant for human disease.
Sources of Funding
This study was supported by the Fritz-Thyssen-Foundation to A. Zirlik and by the German Research Foundation to D. Wolf (WO1994/1-1).
In December 2017, the average time from submission to first decision for all original research papers submitted to Circulation Research was 13.60 days.
The online-only Data Supplement is available with this article at http://circres.ahajournals.org/lookup/suppl/doi:10.1161/CIRCRESAHA.117.312055/-/DC1.
- Nonstandard Abbreviations and Acronyms
- adipocyte triglyceride lipase
- adipose tissue macrophage
- brown adipose tissue
- diet-induced obesity
- epididymal VAT
- free fatty acids
- glucose transporter
- high-fat diet
- hormone-sensitive lipase
- insulin receptor
- tumor necrosis factor
- tumor necrosis receptor–associated factor
- uncoupling protein-1
- visceral adipose tissue
- wild type
- Received September 12, 2017.
- Revision received January 17, 2018.
- Accepted January 19, 2018.
- © 2018 American Heart Association, Inc.
- Bhupathiraju SN,
- Hu FB
- Ortega FB,
- Lavie CJ,
- Blair SN
- Fuster JJ,
- Ouchi N,
- Gokce N,
- Walsh K
- Stanley TL,
- Zanni MV,
- Johnsen S,
- Rasheed S,
- Makimura H,
- Lee H,
- Khor VK,
- Ahima RS,
- Grinspoon SK
- Zirlik A,
- Lutgens E
- Missiou A,
- Köstlin N,
- Varo N,
- et al
- Zhang HH,
- Halbleib M,
- Ahmad F,
- Manganiello VC,
- Greenberg AS
- Schreiber R,
- Diwoky C,
- Schoiswohl G,
- Feiler U,
- Wongsiriroj N,
- Abdellatif M,
- Kolb D,
- Hoeks J,
- Kershaw EE,
- Sedej S,
- Schrauwen P,
- Haemmerle G,
- Zechner R
Novelty and Significance
What Is Known?
Obesity is accompanied by a chronic, low-grade inflammation in adipose tissue.
It is unclear whether increased adipose tissue inflammation directly aggravates obesity and its metabolic complications, such as insulin resistance.
What New Information Does This Article Contribute?
A genetic absence of the inhibitory signaling adaptor TRAF-1 (tumor necrosis receptor–associated factor 1), which limits several inflammatory pathways, is associated with sustained inflammation of mouse adipose tissue.
TRAF-1–deficient mice on high-fat diet showed increased lipolysis in inflamed adipose tissue, enhanced lipid breakdown in adipocytes, and were protected from obesity and insulin resistance.
Inflammatory cytokines released by TRAF-1–deficient adipose tissue macrophages were required to promote lipolytic pathways in TRAF-1–deficient adipocytes.
Human obesity and insulin resistance are associated with inflammation in adipose tissue. Therefore, it has been proposed that inflammatory pathways promote obesity and dysmetabolism. In this study, in a gain-of-function model of inflammation by knocking out the inhibitory anti-inflammatory adapter TRAF-1, we found that enhanced inflammation can synergize with lipolytic pathways to increase the breakdown of lipids in adipocytes, reduce adipose tissue depots, and improve obesity and its metabolic complications. Thus, inflammation of obese adipose tissue is not necessarily detrimental but may instead be required to maintain a beneficial remodeling of adipose tissue. These findings question the overall pathogenic role of inflammation in cardiometabolic disease.