IL-6–Type Cytokine Signaling in Adipocytes Induces Intestinal GLP-1 Secretion

High production and release of interleukin (IL)-6 from adipose tissue may contribute to dysregulated metabolism in obesity and, thus, contribute to the development of insulin resistance (1–3). Consistently, we recently reported a role of IL-6 signaling in adipocytes in the development of obesity-associated liver insulin resistance and steatosis by making use of adipocyte-specific glycoprotein 130 (gp130) knockout (gp130^Δadipo) mice (4). gp130 is a common signal transducer protein of all IL-6–type cytokines that comprises eight different cytokines such as IL-6, IL-11, and leukemia inhibitor factor (LIF) (5). It was previously suggested that these IL-6–type cytokines may affect leptin production in adipose tissue (6–8), thereby potentially mediating a protective effect on glucose metabolism. Interestingly, leptin was found to stimulate the release of glucagon-like peptide 1 (GLP-1) from enteroendocrine cells (9). GLP-1 is an important inducer of glucose-stimulated insulin release in vivo, and it also inhibits gastric emptying and glucagon secretion (10). Hence, identification of factors promoting endogenous GLP-1 release is of interest because enhancing endogenous GLP-1 secretion may be a useful strategy to prevent pancreatic β-cell failure in insulin-resistant obese patients and subsequent development of type 2 diabetes (11). Active GLP-1 is mainly secreted into circulation from intestinal L cells after being cleaved from its precursor proglucagon by the prohormone convertase 1/3 (PC1/3) (10,12,13). The latter is encoded by proprotein convertase subtilisin/kexin type 1 (Pcsk1) that is expressed in intestinal L cells as well as α cells of pacreatic islets (13). Accordingly, GLP-1 can also be produced in pancreatic islets, where it shows paracrine actions (12). In fact, it has recently been proposed that GLP-1 originating from pancreatic islets may be more important for glucose homeostasis than GLP-1 produced in intestinal L cells (14,15).

In the current study, we hypothesized that adipocyte-specific IL-6–type cytokine signaling induces leptin-mediated GLP-1 release in obese mice, thereby improving glucose-stimulated insulin release and glucose tolerance.

Adipocyte-specific gp130 knockout mice (gp130^Δadipo) on a C57BL/6J background were generated by crossing gp130 floxed (gp130^F/F) mice (16) with animals expressing the Cre recombinase controlled by the Adipoq promoter (AdipoqCre mice; purchased from The Jackson Laboratory, Bar Harbor, ME). Six-week-old male mice were fed ad libitum with standard rodent diet (chow) or high-fat diet (HFD; D12331; Research Diets, New Brunswick, NJ) for 12 weeks. HFD consisted of 58% of calories derived from fat, 28% from carbohydrate, and 16% from protein. All protocols conformed to the Swiss animal protection laws and were approved by the Cantonal Veterinary Office in Zurich, Switzerland.

Intraperitoneal glucose and insulin tolerance tests were performed as described (17). To block GLP-1 action, exendin 9–39 (25 nmol/L · kg body weight; Bachem, Bubendorf, Switzerland), was injected intraperitoneally 1 min prior to glucose injection. Blood glucose concentration was measured in blood from tail-tip bleedings using a glucometer (Accu-Chek Aviva; Roche Diagnostics, Rotkreuz, Switzerland).

Glucose (2 g/kg body weight) was orally administered in overnight-fasted mice. To block GLP-1 action, exendin 9–39 (25 nmol/L · kg body weight) was injected intraperitoneally 15 min prior to oral glucose administration. Blood glucose concentration was measured in blood from tail-tip bleedings using a glucometer (Accu-Chek Aviva; Roche Diagnostics).

Food intake was determined using a metabolic and behavioral monitoring system (PhenoMaster; TSE Systems, Bad Homburg, Germany) as described (18).

Adipocytes were isolated as described (19). Isolated adipocytes were incubated in the absence or presence of 1 nmol/L ciliary neurotrophic factor (CNTF; BioLegend, San Diego, CA), 1 nmol/L cardiotrophin-1 (CT-1; BioLegend), 5 nmol/L recombinant IL-6 (AdipoGen), 10 nmol/L IL-11 (Lucerna Chem, Luzern, Switzerland), 2.5 nmol/L LIF (Miltenyi Biotec, Bergisch Gladbach, Germany), or 1 nmol/L oncostatin M (OSM; BioLegend) for 4 h. Leptin levels were measured using MSD technology (Meso Scale Discovery, Gaithersburg, MD).

To prevent GLP-1 degradation, sitagliptin (25 µg/g body weight; Sigma-Aldrich, Buchs, Switzerland) was injected intraperitoneally 30 min before oral or intraperitoneal glucose administration. Of note, sitagliptin was demonstrated to increase circulating GLP-1 levels in mice (20). EDTA (5 mmol/L) and Diprotin A (0.1 mmol/L; Sigma-Aldrich) were added to the collected blood followed by immediate centrifugation at 4°C, and plasma was stored at −80°C until further processing.

Plasma active GLP-1 and leptin were determined using MSD technology. Plasma insulin levels were measured using an ELISA kit as described previously (21). ELISA kits were used to measure plasma adiponectin (Adipoen, Liestal, Switzerland), dipeptidyl peptidase IV (DPP IV; Abnova, Taipei City, Taiwan), and IL-6 (BioLegend).

Total RNA was extracted (NucleoSpin; Macherey-Nagel, Düren, Germany), and concentration was determined spectrophotometrically (Nanodrop 1000; Nanodrop Technologies, Boston, MA). RNA was reverse transcribed with the GoScript Reverse Transcription System (Promega, Madison, WI). TaqMan assays (Applied Biosystems, Rotkreuz, Switzerland) were used for real-time PCR amplification. The following PCR primers (Applied Biosystems) were used: leptin, Mm00434759 m1; and pcsk1, Mm00479023_m1. Relative gene expression was obtained after normalization to 18sRNA (Applied Biosystems) using the formula 2^−ΔΔcp (22).

For mouse islet isolation, pancreata were perfused via the pancreatic duct with a collagenase solution (Worthington Biochemical Corporation, Lakewood, NJ), excised, and digested in the same solution at 37°C for 27 min. The digest was sequentially filtered through 500- and 70-μm cell strainers (BD Falcon; BD Biosciences, Bedford, MA). Islets were handpicked under a microscope and cultured in 24-well extracellular matrix–coated plates (Novamed Ltd., Jerusalem, Israel) for 40 h in RPMI 1640 containing 11.1 mmol/L glucose, 100 units/mL penicillin, 100 μg/mL streptomycin, 2 mmol/L glutamax, 50 μg/mL gentamycin, 10 μg/mL Gibco Fungizone (Thermo Fisher Scientific, Reinach, Switzerland), and 10% FCS at a density of 25 islets/well. To determine glucose-stimulated insulin secretion (GSIS), islets were preincubated for 30 min in modified Krebs-Ringer bicarbonate buffer (KRB) containing 115 mmol/L NaCl, 4.7 mmol/L KCl, 2.6 mmol/L CaCl₂ 2H₂O, 1.2 mmol/L KH₂PO₄, 1.2 mmol/L MgSO₄ 7H₂O, 10 mmol/L HEPES, 0.5% BSA, pH 7.4, and 2.8 mmol/L glucose. KRB was then replaced by KRB 2.8 mmol/L glucose and collected after 1 h (basal insulin release), followed by collection of 1 h release in KRB 16.7 mmol/L glucose (stimulated insulin release) or KRB 16.7 mmol/L glucose with 100 nmol/L GLP-1(7–36) amide (Bachem, Bubendorf, Switzerland). For determination of the insulin content, islet cells were extracted with 0.18 N HCl in 70% EtOH. Insulin concentrations were determined using mouse insulin ultrasensitive mouse/rat insulin kit (Meso Scale Discovery).

Terminal ileum was dissected, flushed in PBS, and cut into smaller pieces of 0.5-cm length. Explants were maintained in culture medium (Krebs buffer containing 0.1% BSA, 5 mmol/L D-glucose, and 100 μmol/L of Diprotin A [Sigma-Aldrich]). After 1 h, medium was changed and supplemented with 15 mmol/L D-glucose or 0.1 μmol/L recombinant leptin (Bio-Techne, Abingdon, U.K.) for 2 h. Thereafter, supernatants were collected, and explants were snap frozen until further processing.

GLUTag cells were grown in DMEM (1 g/L glucose with L-glutamine and pyruvate) supplemented with 10% FBS and 1% penicillin-streptomycin. For experiments, cells were seeded in 12-well (200,000 cells/well) or 24-well (100,000 cells/well) plates. After 1 day, cells were fasted overnight in serum-free medium and subsequently stimulated with or without 0.1 μmol/L recombinant leptin (Bio-Techne) for 6 h. Alternatively, fasted GLUTag cells were treated with supernatants of isolated adipocytes for 6 h in the presence or absence of neutralizing leptin antibody (0.2 µg/mL; Bio-Techne).

Data are presented as means ± SEM. Data were analyzed by unpaired two-tailed Student t test, one-sample t test, one-way ANOVA with Newman-Keuls correction for multiple-group comparisons, or two-way ANOVA with Bonferroni multiple comparisons. All statistical tests were calculated using Prism 5.04 (GraphPad Software, San Diego, CA). P values <0.05 were considered to be statistically significant.

To investigate a possible role of adipocyte-specific IL-6–type cytokine signaling in glucose metabolism, intraperitoneal glucose tolerance tests were performed in adipocyte-specific gp130 knockout mice (gp130^Δadipo) and control littermate mice (gp130^F/F) fed a chow or HFD for 12 weeks. As previously shown, gp130 protein levels are reduced in isolated adipocytes but not in skeletal muscle and liver of gp130^Δadipo mice compared with control littermates, and body weight (4) as well as food intake was similar in both genotypes (Supplementary Fig. 1). As depicted in Fig. 1A and B, depletion of gp130 in adipocytes led to a stronger deterioration of intraperitoneal glucose tolerance in HFD-fed mice. In contrast, intraperitoneal insulin tolerance was improved in HFD-fed gp130^Δadipo mice (Fig. 1C and D), confirming previous findings obtained from hyperinsulinemic-euglycemic clamp studies (4). Such data indicate that impaired glucose tolerance in knockout mice may be the result of blunted GSIS. Indeed, glucose-stimulated circulating insulin levels were significantly lower after intraperitoneal glucose injection in HFD-fed gp130^Δadipo compared with gp130^F/F mice (Fig. 1E). To investigate whether impaired β-cell function may drive the observed phenotype, GSIS was assessed in islets ex vivo. Of note, GSIS was not impaired in islets isolated from HFD-fed gp130^Δadipo mice when compared with control littermates (Fig. 1F). Moreover, insulin content was higher in HFD-fed knockout compared with control mice (Supplementary Fig. 2). Hence, deteriorated glucose-stimulated insulin levels in vivo do not result from defective β-cell function. Rather, impaired incretin secretion may constitute the observed metabolic phenotype.

To assess a potential involvement of impaired incretin secretion, oral glucose tolerance tests were performed. As depicted in Fig. 2A and B, oral glucose tolerance was significantly impaired in HFD-fed knockout mice, paralleled by an ∼40% reduction in circulating insulin levels (Fig. 2C). Of note, the difference between knockout and control mice was more pronounced after oral compared with intraperitoneal glucose administration (∼20% difference in area under the curve [AUC] between HFD-fed gp130^Δadipo and control littermates after oral administration [Fig. 2B] vs. ∼10% difference in AUC after intraperitoneal glucose injection [Fig. 1B]). Because glucose clearance during oral glucose tolerance test is augmented by incretins (23), such data further support a role for decreased incretin secretion as a cause for impaired glucose tolerance in gp130 knockout mice. Indeed, basal as well as glucose-stimulated levels of glucagon-like peptide 1 (GLP-1) were reduced by ∼50% in knockout mice (Fig. 2D). Importantly, the GLP-1 receptor antagonist exendin 9–39 (24) blunted the difference in glucose tolerance between HFD-fed control and knockout mice (Fig. 2E and F). In fact, exendin 9–39 significantly blunted AUC in gp130^F/F mice (1,272 ± 57 mmol/L · min in control vs. 1,500 ± 51 mmol/L · min in exendin 9–39-treated mice; P < 0.05) but not in gp130^Δadipo mice (1,507 ± 34 mmol/L · min in control vs. 1,567 ± 34 mmol/L · min in exendin 9–39-treated mice; P = 0.23). Besides affecting oral glucose tolerance, exendin 9–39 treatment abrogated impaired intraperitoneal glucose tolerance in gp130^Δadipo mice (Fig. 2G), suggesting that reduced basal GLP-1 levels (Fig. 2D) contributed to impaired intraperitoneal glucose tolerance in untreated mice (Fig. 1A and B). Of note, ex vivo stimulation of pancreatic islets with high glucose concentration in combination with GLP-1 revealed similar degrees of insulin secretion in HFD-fed gp130^F/F and gp130^Δadipo mice (Fig. 2H), indicating similar GLP-1 sensitivity of β-cells. Taken together, adipocyte-specific depletion of IL-6–type cytokine signaling reduces GLP-1 release in HFD-fed mice, thereby impairing glucose tolerance.

Reduced circulating GLP-1 levels in HFD-fed gp130^Δadipo mice may result from enhanced degradation by DPP IV that is increasingly released from adipose tissue in obesity (10,25). However, plasma DPP IV levels were similar in HFD-fed gp130^F/F and gp130^Δadipo mice (Supplementary Fig. 3), indicating that IL-6–type cytokine signaling did not affect circulating DPP IV levels and, hence, was not responsible for blunted GLP-1 levels in knockout mice. Circulating GLP-1 derives almost completely from enteroendocrine L cells (12). Thus, reduced circulating GLP-1 levels in gp130 knockout mice may result from blunted intestinal secretion. Indeed, glucose-stimulated GLP-1 secretion from intestinal explants harvested from HFD-fed gp130^Δadipo mice was reduced (Fig. 3A), whereas its release from islets isolated from HFD-fed knockout mice remained unaffected (Supplementary Fig. 4). Hence, reduced circulating GLP-1 levels in HFD-fed gp130^Δadipo mice resulted from its blunted release from enteroendocrine cells.

Recently, different adipose-derived circulating factors were shown to affect GLP-1 release from intestinal cells. In fact, circulating IL-6 was shown to stimulate GLP-1 secretion, thereby enhancing GSIS (26). Moreover, the two adipokines adiponectin and leptin were reported to induce GLP-1 secretion from intestinal L cells (9,27). As depicted in Supplementary Fig. 5, circulating IL-6 and adiponectin levels were similar between HFD-fed gp130^F/F and gp130^Δadipo mice. In contrast, plasma leptin levels were reduced by ∼40% in gp130^Δadipo mice (Fig. 3B) despite similar body weight (4), potentially contributing to their reduced GLP-1 levels. To analyze the impact of IL-6–type cytokine signaling on leptin production in adipocytes, its expression and release were analyzed in isolated adipocytes of chow- and HFD-fed knockout and control mice. As expected, HFD induced leptin mRNA expression in and release from adipocytes of control mice (Fig. 3C and D). Importantly, leptin release from adipocytes was significantly blunted in HFD-fed knockout mice (Fig. 3D), whereas its expression was unchanged (Fig. 3C). In addition, leptin content in isolated adipocytes of HFD-fed mice was similar between the genotypes (Fig. 3E). Hence, IL-6–type cytokine signaling in adipocytes affects leptin release but not its transcription and protein synthesis. We next wanted to investigate which member of the IL-6–type cytokine family mediated the effect on leptin release. To this end, isolated adipocytes of chow-fed gp130^F/F and gp130^Δadipo mice were incubated with the IL-6–type cytokine family members CNTF, CT-1, IL-6, IL-11, LIF, or OSM, which were previously suggested to affect leptin production or were suggested as potential targets to treat obesity (5–8,28). Although CNTF, CT-1, IL-11, LIF, and OSM had no significant effect on leptin release, IL-6 stimulated leptin secretion in isolated adipocytes gp130 dependently (Fig. 3F). Taken together, IL-6 induced leptin release from adipocytes, thereby contributing to elevated circulating leptin levels in HFD-fed mice.

Leptin induces GLP-1 release from the enteroendocrine cell line GLUTag cells (9,29). Such release of GLP-1 is regulated by PC1/3 in intestinal cells (10,13). Indeed, expression of Pcsk1, the gene encoding PC1/3 (13), was significantly increased by leptin treatment (Fig. 4A). Importantly, leptin also stimulated Pcsk1 expression in L-cell–rich ileum (14) explants harvested from HFD-fed gp130^F/F mice (Fig. 4B). Next, experiments in GLUTag cells that were incubated with supernatant harvested from isolated adipocytes were performed to further explore the proposed adipo-enteroendocrine axis driven by leptin. As shown in Fig. 4C, GLP-1 release was significantly blunted in GLUTag cells incubated with supernatants collected from adipocytes of HFD-fed gp130^Δadipo mice. Of note, leptin concentration was reduced in such supernatants, as shown previously (Fig. 3D). Importantly, leptin neutralization reversed observed difference on GLP-1 release in GLUTag cells incubated with supernatant collected from adipocytes (Fig. 4D), suggesting a critical role of leptin in the observed effect. In parallel, Pcsk1 expression was reduced (Fig. 4E) in GLUTag cells incubated with supernatants collected from adipocytes of HFD-fed gp130^Δadipo mice, an effect that was reversed upon leptin neutralization (Fig. 4F). Likewise, expression of Pcsk1 was reduced in the ileum of HFD-fed knockout mice (Fig. 4G). Collectively, our data suggest that IL-6–type cytokine signaling in adipocytes promotes intestinal GLP-1 release leptin dependently, probably by stimulating Pcsk1 expression.

The current study suggests that adipocyte-specific IL-6–type cytokine signaling promotes leptin-mediated GLP-1 release from enteroendocrine cells in obesity, thereby improving glucose-stimulated insulin release and glucose tolerance (Fig. 5). Such a notion is based on the following findings: 1) gp130 depletion specifically in adipocytes impaired glucose tolerance and reduced circulating leptin, GLP-1, and insulin levels in HFD-fed mice; and 2) GLP-1 secretion is reduced leptin dependently from enteroendocrine cells incubated with supernatant collected in gp130-depleted adipocytes.

Reduced GLP-1 release from enteroendocrine cells in adipocyte-specific gp130 knockout mice may result from reduced Pcsk1 expression, the gene encoding PC1/3 controlling GLP-1 production (10,13). Leptin was previously reported to enhance Pcsk1 expression in neuronal cells (30). In HFD-fed gp130^Δadipo mice, intestinal Pcsk1 expression was lower compared with control littermates and associated with significantly lower circulating leptin levels. Moreover, supernatant collected from isolated adipocytes regulated Pcsk1 expression in GLUTag cells in a leptin-dependent manner, and recombinant leptin stimulated Pcsk1 expression in GLUTag cells as well as in ileum explants of HFD-fed mice. Of note, the role of intestinal GLP-1 in glucose homeostasis has recently been questioned, suggesting that pancreatic GLP-1 may be more important for glucose homeostasis (14). Indeed, we did not find diminished GLP-1 release from islets isolated from HFD-fed knockout mice. Hence, our data indicate that reduced intestinal rather than pancreatic GLP-1 release affects GSIS and, consequently, glucose homeostasis in gp130^Δadipo mice. Although we cannot rule out a role of pancreatic GLP-1, our data clearly support an important role of intestinal GLP-1 production in glucose homeostasis. In support of an involvement of the incretin system in the observed metabolic phenotype in HFD-fed gp130^Δadipo mice, insulin release and content were not reduced in islets of HFD-fed knockout mice ex vivo. In fact, insulin content was rather elevated in islets of gp130^Δadipo mice, potentially mirroring reduced (GLP-1–stimulated) insulin secretion in vivo. GLP-1 sensitivity of isolated islets was similar between HFD-fed control and knockout mice, suggesting that reduced circulating GLP-1 levels rather than GLP-1 resistance of β-cells cause impaired insulin secretion in knockout mice. In support of such notion, administration of oral glucose, which induces circulating GLP-1 levels in contrast to intraperitoneal glucose injection, exacerbated the difference in glucose tolerance between knockout and control mice.

Lack of IL-6–type cytokine signaling in adipocytes blunts circulating leptin levels independently of fat pad mass (4). Interestingly, reduced circulating leptin levels in knockout mice were paralleled by reduced leptin release from but similar mRNA expression in adipocytes, indicating that gp130 depletion affects leptin secretion but not leptin transcription. Accordingly, it has been suggested that circulating signals regulate leptin on a posttranscriptional level, whereas the long-term nutritional status affects leptin mRNA expression (31). In the past, several IL-6–type cytokines were suggested to affect leptin production or proposed as potential targets to treat obesity (5–8,28). In this study, we suggest that IL-6 contributes to obesity-induced leptin release from adipocytes, adding another metabolic function to the pleiotropic actions of IL-6 (32–34). In fact, among all tested IL-6–type cytokines, only IL-6 significantly increased leptin release from isolated adipocytes. Consistently, treatment of HFD-fed C57BL/6 mice with an anti–IL-6 receptor antibody reduced circulating leptin levels (35). Clearly, we cannot rule out that other IL-6–type family members and/or other molecules binding gp130, such as superantigens (36), contribute to reduced circulating leptin levels in knockout mice. In addition, compensatory changes in the production of other adipose-derived factors in knockout mice may have contributed to reduced leptin levels.

Despite impaired glucose tolerance, insulin sensitivity was improved in HFD-fed gp130^Δadipo mice. This observation may be explained by reduced portal free fatty acid (FFA) flux in knockout mice, leading to blunted hepatic insulin resistance and steatosis (4). Hence, IL-6–type cytokine signaling in adipocytes may promote both protective as well as harmful effects on glucose metabolism (Fig. 5). Physiologically, increased IL-6–mediated release of FFA and leptin from adipocytes in obesity may be interpreted as an attempt to reduce lipid stores by increasing lipolysis (FFA release) and by lowering lipid accumulation via leptin-mediated reduction of food intake. Notably, the observed leptin-mediated effect of IL-6 on circulating GLP-1 levels may be additive to the direct effect of IL-6 on GLP-1 release from enteroendocrine cells (26).

Of note, reduced circulating leptin levels in knockout mice did not significantly affect body weight (4), suggesting similar central leptin action between control and knockout mice. In agreement, food intake was similar between HFD-fed gp130^F/F and gp130^Δadipo littermates. Such a finding may suggest that HFD feeding induces a central leptin resistance, whereas leptin-mediated signaling in the periphery (e.g., in intestinal cells) is less affected (37,38). Accordingly, leptin induced Pcsk1 expression in the ileum of HFD-fed mice. Alternatively, decreased leptin passage across the blood–brain barrier in obese mice and/or local GLP-1 production in the brain (10,37) may lead to unaffected GLP-1 levels in the brain of knockout mice. Clearly, further studies are needed to shed more light on involved mechanisms. In addition, adipocyte-specific depletion of gp130 in female mice may result in a different metabolic phenotype, as leptin secretion and the potency of GLP-1-function are sex dependent (39).

In conclusion, we identify a novel adipo-enteroendocrine axis driven by IL-6 in the regulation of glucose-stimulated insulin release and glucose tolerance in obesity. Such an axis may sensitize pancreatic β-cells to glucose and, thus, counteract adipose tissue-induced insulin resistance (4,40). Moreover, it may offset carbohydrate and/or leptin resistance of enteroendocrine cells, resulting in augmented GLP-1 levels in obesity (9,41–43).

Funding. This work was supported by research grants from the Swiss National Science Foundation (310030-160129) and the Gottfried und Julia Bangerter-Rhyner-Stiftung (both to D.K.).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. S.W. and D.K. conceived the study and wrote the manuscript. S.W., C.I.L., M.B.-S., F.I., and F.C.L. performed the experimental work. W.M. provided gp130^Δadipo mice and gave conceptual advice. M.Y.D. gave conceptual advice. S.W., C.I.L., M.B.-S., F.I., F.C.L., M.B., W.M., M.Y.D., and D.K. contributed to discussion and reviewed and edited the manuscript. D.K. is the guarantor of this work and, as such, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the 77th Scientific Sessions of the American Diabetes Association, San Diego, CA, 9–13 June 2017.