Triggering receptor expressed on myeloid cells 2 (TREM2) is highly expressed by innate immune cells and has been implicated in a variety of diseases, including Alzheimer disease, cancer, and diabetes. TREM2 binds a wide array of ligands signaling tissue damage (1), including pathogenic lipids, which can elicit highly variable effects (both in terms of magnitude and direction) on downstream signaling pathways in a ligand-specific manner (2). Given the importance of pathogenic lipids and myeloid cells in obesity and diabetes, there has been considerable interest in elucidating the role of TREM2 in disease progression. Metabolic studies in mice globally overexpressing TREM2 (3) or in mice with global TREM2 knockout (Trem2−/−) (4,5) both reported increased body weight gain and exacerbated insulin resistance. Thus, the relationship between TREM2 and insulin resistance has proven complicated, and the mechanism(s) underlying its effects on metabolic disease require further elucidation to help develop a therapeutic approach.
In this issue of Diabetes, Sharif et al. (6) identify ceramides, a class of sphingolipids, as an important link between TREM2 and insulin resistance (measured by the insulin tolerance test). Ceramides play a role in membrane integrity, cellular stress responses, inflammation, and apoptotic signaling, and elevated ceramide levels have been implicated in numerous metabolic pathologies (7). Using a metabolomics approach, the authors demonstrate elevated serum ceramide levels in Trem2−/− mice at baseline and during diet-induced obesity (DIO) relative to wild-type (WT) mice. Strikingly, inhibition of de novo ceramide synthesis using myriocin lowered circulating ceramides and improved insulin sensitivity in Trem2−/− mice during DIO independent of body weight gain or adiposity.
Adipose tissue macrophages (ATMs) expand during obesity (8), bringing about chronic, low-grade inflammation while also clearing dead adipocytes in white adipose tissue (WAT) (9,10). Their ability to uptake excess lipids as well as the expansion of TREM2+ ATMs during obesity position them as likely mediators of the metabolic effects of Trem2−/−. In the current study, Trem2−/− mice lacked the expansion of monocyte-derived “FBC” (F4/80+CD11b+ CD11c+) ATMs observed in WT mice, but Trem2−/− did not change overall ATM number or their inflammatory cytokine expression after 13 weeks of high-fat diet (HFD). These Trem2−/− mice did, however, display lower ATM lipid loads. Notably, inhibiting de novo ceramide synthesis with myriocin restored insulin sensitivity without significantly impacting any of these ATM parameters, suggesting that the effects of TREM2 effects on insulin resistance may be macrophage independent (Fig. 1).
Sharif et al. (6) instead propose that TREM2’s role in insulin resistance may be intrinsic to WAT itself. They detect adipose tissue expression of TREM2 not only in ATMs but also in adipocytes, suggesting that insulin resistance may be impacted by TREM2 expression by both cell types. This scenario draws an intriguing parallel to earlier work on fatty acid binding protein 4 (FABP4), whose effects on metabolism required cell-specific dissection due to its expression by both ATMs and adipocytes (11,12). In addition to having elevated circulating ceramides, Trem2−/− mice on prolonged HFD had elevated liver ceramides and exacerbated hepatic steatosis, further suggesting a defect in lipid storage and/or metabolism in the absence of TREM2. Prolonged HFD feeding also results in ATM loss over time in Trem2−/− mice, and, intriguingly, transplant of Trem2+/+ bone marrow cells failed to restore ATMs, insulin sensitivity, and hepatic lipid levels. This led the authors to hypothesize that Trem2−/− decreased chemotactic signaling in WAT. Indeed, compared with WT, media conditioned by Trem2−/− adipocytes contained dramatically less MCP-1, a key macrophage chemoattractant. Overall, these studies provide evidence that nonhematopoietic TREM2 influences the metabolic phenotype of Trem2−/− mice during DIO.
A major advance of the work by Sharif et al. (6) is the demonstration that elevated ceramides contribute to insulin resistance in obese Trem2−/− mice. Interestingly, the authors’ metabolomics approach also uncovered an increase in serum ceramides in lean Trem2−/− mice, suggesting a predisposition of these animals to poor metabolic outcomes in response to HFD and providing a glimpse into TREM2’s potential mechanism of action in maintaining insulin sensitivity. The role of TREM2 in lipid homeostasis fits with its function in Alzheimer disease, where it upregulates lipid metabolism gene expression in disease-associated microglia (13).
Another important aspect of this study is that inhibition of de novo ceramide synthesis improved insulin sensitivity without impacting ATM numbers or inflammation in Trem2−/− mice. While this could indicate that TREM2’s influence on insulin resistance is driven by ceramides and independent of macrophages, this interpretation has limitations. First, ceramide production by macrophages themselves can be stimulated by saturated fatty acids, which are elevated during obesity (14), and these ceramides have been previously shown to contribute to insulin resistance (15). Thus, the data presented do not negate the possibility that macrophage-derived ceramides (not measured in this study) contribute to insulin resistance in Trem2−/− mice, which may be important to assess in future studies.
Second, while Sharif et al. (6) suggest that TREM2’s effects on metabolism are macrophage independent, recent studies by Jaitin et al. (4) suggest the opposite. These discordant findings may be explained in part by the utilization of bone marrow transplantation (BMT) to discern the importance of hematopoietic cells. While BMTs are commonly used in obesity studies, the irradiation and subsequent antibiotic treatment often implemented can confound interpretation due to damage to intestinal epithelia and loss of gut microbial diversity (16–18). Moreover, dynamics of immune cell ablation by radiation and reconstitution posttransplant vary by cell type and tissue niche (19). This may be particularly important for macrophages, whose activities in multiple tissues contribute to insulin resistance (20). As Sharif et al. demonstrate, BMT with Trem2+/+ bone marrow did not ameliorate ATM loss following prolonged HFD in Trem2−/− mice despite successful reconstitution of blood monocytes. These findings evoke the intriguing question of whether Trem2+/+ macrophages would restore insulin sensitivity if they were better able to infiltrate Trem2−/− adipose tissue. Implementing Cre technology, which overcomes many of the limitations of BMT, may help to clarify the role of macrophage TREM2 in metabolic disease.
Overall, the work presented by Sharif et al. advances our understanding of the role of TREM2 in the development of insulin resistance via its influences on adipose tissue signaling and sphingolipid metabolism. Future studies in adipocyte- and macrophage-specific TREM2 knockouts in obesity and diabetes as well as in other TREM2-dependent disease models will prove instrumental in leveraging this sensor of tissue damage toward therapeutic interventions.
See accompanying article, p. 2042.
Duality of Interest. No potential conflicts of interest relevant to this article were reported.