Dipeptidyl peptidase 4 (DPP-4) cleaves a large number of chemokine and peptide hormones involved in the regulation of the immune system. Additionally, DPP-4 may also be involved in macrophage-mediated inflammation and insulin resistance. Thus, the current study investigated the effect of linagliptin, an inhibitor of DPP-4, on macrophage migration and polarization in white adipose tissue (WAT) and liver of high-fat diet–induced obese (DIO) mice. DPP-4+ macrophages in lean and obese mice were quantified by fluorescence-activated cell sorting (FACS) analysis. DPP-4 was predominantly expressed in F4/80+ macrophages in crown-like structures compared with adipocytes in WAT of DIO mice. FACS analysis also revealed that, compared with chow-fed mice, DIO mice exhibited a significant increase in DPP-4+ expression in cells within adipose tissue macrophages (ATMs), particularly M1 ATMs. Linagliptin showed a greater DPP-4 inhibition and antioxidative capacity than sitagliptin and reduced M1-polarized macrophage migration while inducing an M2-dominant shift of macrophages within WAT and liver, thereby attenuating obesity-induced inflammation and insulin resistance. Loss of macrophage inflammatory protein-1α, a chemokine and DPP-4 substrate, in DIO mice abrogated M2 macrophage-polarizing and insulin-sensitizing effects of linagliptin. Therefore, the inhibition of DPP-4 by linagliptin reduced obesity-related insulin resistance and inflammation by regulating M1/M2 macrophage status.
Obesity activates the innate immune system with subsequent recruitment of immune cells such as macrophages and T cells into metabolic tissues, leading to the development of insulin resistance (1). Macrophage recruitment and polarization are pivotal in obesity-induced inflammation and insulin resistance (2–5). However, treatment options that target immune cells with the aim of halting the development of insulin resistance and type 2 diabetes remain limited.
Dipeptidyl peptidase 4 (DPP-4) inhibitors are effective in the treatment of type 2 diabetes, as they maintain blood glucose levels through degradation of incretin peptides, glucagon-like peptide 1 (GLP-1) and glucose-dependent insulinotropic polypeptide (6). Numerous investigations have detailed the effects of DPP-4 inhibitors on insulin and/or glucagon secretion, but little evidence indicates that DPP-4 inhibitors directly improve chronic inflammation. Although several animal studies suggest that DPP-4 inhibition may attenuate obesity-associated inflammation and cardiovascular disease (7,8), mechanisms describing these actions have yet to be elicited.
DPP-4, originally known as T-cell surface marker CD26, is widely expressed in many cells, including immune cells (9). DPP-4 cleaves a large number of chemokine and peptide hormones involved in the regulation of the immune system, inferring a role for DPP-4 in the pathogenesis of inflammation (6,10). Earlier work has focused on the function of DPP-4 as a surface protease involved in T cell activation (11). Recent studies show that DPP-4 release correlates with adipocyte size or visceral adiposity, and thus, DPP-4 has come to be considered as an adipokine with potential to impair insulin sensitivity (12,13). In addition, DPP-4 on dendritic cell/macrophages contributes to potentiating inflammation of adipose tissue in obesity (14). However, the role of DPP-4 in macrophage-mediated inflammation and insulin resistance remains largely unknown.
Linagliptin is a DPP-4 inhibitor with a high affinity for DPP-4 in various tissues (15–17). In rodents, linagliptin more effectively reduces DPP-4 activity in tissue and attenuates inflammatory properties under conditions of vascular dysfunction than does sitagliptin (18,19). Therefore, we hypothesized that DPP-4 plays a role in regulating macrophage activation in response to obesity and that DPP-4 inhibition may attenuate obesity-induced inflammation. In the current study, we demonstrated that DPP-4 is predominantly expressed in M1-polarized macrophages in white adipose tissue (WAT) of high-fat (HF) diet–induced obese (DIO) mice. Furthermore, we present evidence suggesting that DPP-4 inhibition attenuates obesity-related insulin resistance and inflammation by regulating both macrophage recruitment and M1/M2 status in DIO mice.
Research Design and Methods
Mice and Diets
Eight-week-old male C57BL/6J mice (Charles River Laboratories, Yokohama, Japan) were divided into four groups and fed for 8 weeks as follows: 1) normal chow (NC) with 10% of calories from fat (CRF-1) (Charles River Laboratories); 2) NC containing 0.003% linagliptin (NC+Lina) (Boehringer Ingelheim Pharma GmbH & Co. KG, Biberach an der Riss, Germany); 3) HF diet, consisting of 60% fat (Research Diets, New Brunswick, NJ); and 4) HF diet with 0.003% linagliptin (HF+Lina). To compare the effects of linagliptin and sitagliptin on DIO mice, groups of 8-week-old C57BL/6J mice were fed with NC, HF, HF+Lina, or an HF diet with 0.01% sitagliptin (HF+Sita) (Sequoia, Oxford, U.K.) for 14 weeks. MIP-1α−/− mice were provided by N. Mukaida (Kanazawa University, Kanazawa, Japan) (20). Eight-week-old male MIP-1α−/− mice were fed with the HF or HF+Lina for 8 weeks. All animal procedures were performed in accordance with the standards set forth in the Guidelines for the Care and Use of Laboratory Animals at Kanazawa University.
DPP-4 Activity and GLP-1 Determinations
C57BL/6J mice were dosed with vehicle, linagliptin (3 mg/kg), or sitagliptin (10 mg/kg) by gavage, once daily for 5 days. Blood was collected before or 2 h and 24 h after drug administration. Tissue and plasma DPP-4 activities were detected with DPP-4 assay kit (BioVision, Milpitas, CA). Plasma active GLP-1 concentration was determined using enzyme-linked immunosorbent assay ELISA (Linco Research, St. Charles, MO).
Histological and Immunofluorescence Staining
Paraffin wax–embedded epididymal WAT (eWAT) and liver sections were stained with hematoxylin and eosin (H&E) and immunohistochemically stained for F4/80, as described previously (5,21). For immunofluorescence staining, epididymal fat pads were stained with perilipin (Sigma-Aldrich, Tokyo, Japan), F4/80 (Abcam, Cambridge, U.K.), CD11c (AbD Serotec, Hercules, CA), CD206 (AbD Serotec), and DPP-4 (R&D Systems, Minneapolis, MN), followed by secondary antibodies (Alexa Fluor 488, Alexa Fluor 594, Cy3, and CF 594; Jackson ImmunoResearch Laboratories, West Grove, PA) using standard techniques.
Fluorescence-Activated Cell Sorting Analysis
Stromal vascular fraction (SVF) and nonparenchymal cells were isolated as described previously (5,21,22) and incubated with Fc-Block (BD Bioscience), followed by incubation with fluorochrome-conjugated antibodies (Supplementary Table 1). Cells were analyzed using a FACSAria II (BD Biosciences, San Jose, CA). Data analysis and compensation were performed using FlowJo (Tree Star, Ashland, OR).
Lipid, Glucose, and Insulin Determinations
Plasma triglycerides (TG), total cholesterol (TC), nonesterified fatty acids (NEFA), aspartate aminotransferase (AST), alanine aminotransferase (ALT), glucose, insulin levels and hepatic TG, thiobarbituric acid reactive substrates (TBARS) concentrations, and urine 8-hydroxy-2′-deoxyguanosine (8-OHdG) content were measured as described previously (21,22). Glucose tolerance test (GTT) was conducted after an overnight fast, and then mice were injected with 2 g/kg i.p. glucose. Insulin tolerance test (ITT) was performed after a 4-h fast, and mice were injected with 0.5 units/kg i.p. human insulin.
Quantitative Real-Time PCR
Quantitative real-time PCR was performed as described previously (21,22). Primers used for real-time PCR are shown in Supplementary Table 2.
Immunoblots
Immunoblots were conducted as described previously (5,21). Antibodies are shown in Supplementary Table 3.
Culture of Peritoneal Macrophages
Peritoneal macrophages were isolated and cultured as described previously (5,21). After starving for 6 h, the cells were coincubated with 100 ng/mL lipopolysaccharide (LPS) (Sigma-Aldrich) or 10 ng/mL interleukin (IL)-4 (Sigma-Aldrich) and DPP-4 (100, 200, and 500 ng/mL) or linagliptin (50, 100, or 200 nmol/L) for 16 h and then harvested. Intracellular reactive oxygen species (ROS) formation was determined by 5-(and-6)-chloromethyl-2′,7′-dichlorodihydrofluorescein diacetateacetylester fluorescent probe as described previously (23).
Statistical Analysis
All data are presented as means ± SEM. Differences between the mean values of two groups were assessed using a two-tailed Student t test. Differences in mean values between more than two groups were determined using ANOVA. A P value <0.05 was considered significant.
Results
Accumulation of DPP-4+ Macrophages Increases in Fat of DIO mice
DPP-4 activity in both the plasma and eWAT significantly increased in HF-fed mice compared with NC-fed mice (Fig. 1A). Expression of mRNA for Dpp-4 was higher in the SVF than adipocyte fraction from both NC and HF-fed mice (Fig. 1B). Moreover, Dpp-4 expression in the SVF in HF-fed mice was markedly higher than that in NC-fed mice.
DPP-4+ ATMs increase in HF diet–induced obesity. C57BL/6J mice fed with NC or HF diet for 8 weeks. A: DPP-4 activity in plasma and eWAT. B: DPP-4 mRNA expression in adipocyte fraction and SVF. C: Immunofluorescence staining for F4/80, CD11c, CD206, perilipin (green), and DPP-4 (red) in eWAT from DIO mice. Scale bars, 50 μm. D: Quantification of DPP-4+ ATMs and DPP-4+ M1 or M2 ATMs in eWAT from NC diet– or HF diet–fed mice by FACS analysis. Gating strategies to determine ATMs are depicted in Supplementary Fig. 1, and fluorescence-minus-one (FMO) controls were used for gating highly pure populations of DPP-4+ macrophages. Black arrows indicate DPP-4+ macrophages with high fluorescent intensity. E: Relative abundance of DPP-4− M1 or DPP-4− M2 and DPP-4+ M1 or DPP-4+ M2 ATMs in eWAT. n = 6–8. *P < 0.05, **P < 0.01 vs. mice fed the NC diet.
DPP-4+ ATMs increase in HF diet–induced obesity. C57BL/6J mice fed with NC or HF diet for 8 weeks. A: DPP-4 activity in plasma and eWAT. B: DPP-4 mRNA expression in adipocyte fraction and SVF. C: Immunofluorescence staining for F4/80, CD11c, CD206, perilipin (green), and DPP-4 (red) in eWAT from DIO mice. Scale bars, 50 μm. D: Quantification of DPP-4+ ATMs and DPP-4+ M1 or M2 ATMs in eWAT from NC diet– or HF diet–fed mice by FACS analysis. Gating strategies to determine ATMs are depicted in Supplementary Fig. 1, and fluorescence-minus-one (FMO) controls were used for gating highly pure populations of DPP-4+ macrophages. Black arrows indicate DPP-4+ macrophages with high fluorescent intensity. E: Relative abundance of DPP-4− M1 or DPP-4− M2 and DPP-4+ M1 or DPP-4+ M2 ATMs in eWAT. n = 6–8. *P < 0.05, **P < 0.01 vs. mice fed the NC diet.
Immunofluorescence analysis of eWAT in HF-fed mice revealed that DPP-4 was expressed by F4/80+ macrophages in crown-like structures, but it was poorly expressed by perilipin+ adipocytes (Fig. 1C). Interestingly, most of CD11c+ M1 macrophages expressed DPP-4, whereas fewer CD206+ M2 macrophages expressed DPP-4 (Fig. 1C). Fluorescence-activated cell sorting (FACS) analysis revealed that the total number of adipose tissue macrophages (ATMs) was 3.3-fold higher in HF-fed mice than the NC-fed mice (Supplementary Fig. 1). When gated for DPP-4/CD26, ATMs exhibited a higher level of DPP-4 expression in response to HF diet (Fig. 1D). Furthermore, HF-fed mice exhibited a significant increase in the percentage of DPP-4+ cells within CD11c+CD206− (M1) ATMs compared with NC-fed mice. However, the percentage of DPP-4+ cells within CD11c−CD206+ (M2) ATMs was unaffected by HF feeding (Fig. 1D). Among ATMs, both DPP-4− M1 and DPP4+ M1 ATMs increased with HF feeding, and yet DPP-4− M2 ATMs decreased, with the expression of DPP4+ M2 ATMs remaining unchanged (Fig. 1E). Collectively, these results support the proposition that DPP-4+ macrophages infiltrate the WAT of obese mice and that DPP-4+ M1 ATMs accumulate in response to obesity.
DPP-4 Regulates M1/M2 Polarization in Macrophages
Given the association between ATM accumulation and DPP-4 activation in obese mice, we next determined whether DPP-4 directly regulates macrophage activation and/or polarization. Exposure of cultured peritoneal macrophages to DPP-4 increased mRNA expression for M1 markers (TNFα, MCP-1, and RANTES) and intracellular ROS production upon stimulating with LPS in a dose-dependent fashion (Fig. 2A). The M1 polarization and increased ROS generation in macrophages were diminished in the presence of linagliptin (Fig. 2B). In contrast, exposure of macrophages to DPP-4 reduced M2 markers of mRNA expression (Arg1, Chi3l3, and Mgl2) and increased intracellular ROS production in a dose-dependent manner when incubated with IL-4 (Fig. 2C). Exposure to linagliptin increased expression of M2 macrophage markers and decreased ROS generation (Fig. 2D). DPP-4 alone had no effect on mRNA expression of macrophage markers or ROS production (Supplementary Fig. 2A and B). Additionally, linagliptin (50–200 nmol/L) decreased the expression of LPS-induced M1 markers mRNA expression and ROS generation. In contrast, linagliptin augmented IL-4–induced M2 marker expression and decreased ROS production in a dose-dependent manner (Supplementary Fig. 2C and D). Taken together, these results suggest that DPP-4 directly regulates M1/M2 macrophage polarization following LPS or IL-4 stimulation, and linagliptin can suppress M1-polarized activation and induce M2-polarized activation, at least in part, by regulating intracellular ROS generation.
DPP-4 enhances M1 polarization and inhibits M2 polarization in macrophages in an ROS-dependent manner. A: DPP-4 upregulates LPS-induced M1 marker mRNA expression and increases intracellular ROS production in peritoneal macrophages. B: Linagliptin (Lina) suppresses the increase of DPP-4–induced (500 ng/mL) M1 marker mRNA expression and ROS production by LPS stimulation. C: DPP-4 inhibits IL-4–induced M2 marker mRNA expression and increases ROS in M2-polarized macrophages. D: Linagliptin restores the decrease of DPP-4–induced (500 ng/mL) M2 marker mRNA expression and decreases ROS production in the presence of IL-4. Peritoneal macrophages were isolated and coincubated with LPS (100 ng/mL) or IL-4 (10 ng/mL) and DPP-4 (100–500 ng/mL) or linagliptin (200 nmol/L) for 16 h. n = 6. *P < 0.05, **P < 0.01 vs. control incubations; †P < 0.05, ††P < 0.01 vs. LPS- or IL-4–stimulated incubations; ‡P < 0.05, ‡‡P < 0.01 vs. LPS/IL-4 and DPP-4 and/or linagliptin-stimulated incubations.
DPP-4 enhances M1 polarization and inhibits M2 polarization in macrophages in an ROS-dependent manner. A: DPP-4 upregulates LPS-induced M1 marker mRNA expression and increases intracellular ROS production in peritoneal macrophages. B: Linagliptin (Lina) suppresses the increase of DPP-4–induced (500 ng/mL) M1 marker mRNA expression and ROS production by LPS stimulation. C: DPP-4 inhibits IL-4–induced M2 marker mRNA expression and increases ROS in M2-polarized macrophages. D: Linagliptin restores the decrease of DPP-4–induced (500 ng/mL) M2 marker mRNA expression and decreases ROS production in the presence of IL-4. Peritoneal macrophages were isolated and coincubated with LPS (100 ng/mL) or IL-4 (10 ng/mL) and DPP-4 (100–500 ng/mL) or linagliptin (200 nmol/L) for 16 h. n = 6. *P < 0.05, **P < 0.01 vs. control incubations; †P < 0.05, ††P < 0.01 vs. LPS- or IL-4–stimulated incubations; ‡P < 0.05, ‡‡P < 0.01 vs. LPS/IL-4 and DPP-4 and/or linagliptin-stimulated incubations.
Linagliptin Ameliorates HF Diet–Induced Insulin Resistance in a Dose-Dependent Manner
To determine the effective doses of linagliptin, C57BL/6J mice were fed the NC, HF, or HF diets containing 0.003% or 0.01% linagliptin for 8 weeks. Administrating linagliptin did not affect weight or adiposity in HF-fed mice (Supplementary Table 4 and Supplementary Fig. 3A). Treatment with linagliptin decreased plasma TG, TC, AST, and ALT levels in a dose-dependent fashion and tended to decrease plasma NEFA levels in HF mice (Supplementary Table 4). Increased plasma DPP-4 activity in HF-fed mice was inhibited by 89.8% and 98.4% in the two doses used, respectively (Supplementary Fig. 3B). Evidence of HF diet–induced glucose intolerance, insulin resistance, and hyperinsulinemia appeared to follow a dose-dependent pattern of inhibition by linagliptin (Supplementary Fig. 3C–E). Greater than 80% inhibition of DPP-4 is effective, as the DPP-4 inhibitor class (24) and insulin-sensitizing effect were observed in the 0.003% linagliptin-treated group; therefore, subsequent experiments were performed using that dose (named HF+Lina).
Linagliptin Protects Mice From HF Diet–Induced Impaired Glucose Homeostasis and Hepatic Steatosis
NC and NC+Lina mice had similar body, liver, and eWAT weights as did the HF and HF+Lina mice (Fig. 3A). However, the histological analysis revealed severe lipid accumulation, which was decreased markedly by linagliptin, in the livers of mice fed the HF diet (Fig. 3B). Linagliptin consistently reduced liver TG in the HF-fed mice (Fig. 3C). These findings were associated with the suppression of lipogenic genes expression (SREBP-1c, FAS, and SCD1) and upregulation of mitochondrial fatty acid β-oxidation genes (PPARα, Cpt1α, and LCAD) (Fig. 3D).
Linagliptin alleviates diet-induced insulin resistance and hepatic steatosis. A: Weight gain and tissue weights of C57BL/6J mice fed an NC or HF diet without or with 0.003% linagliptin (Lina) from age 8 to 16 weeks. B: H&E-stained liver sections from mice. Scale bars, 100 μm. C: Liver TG content. D: mRNA expression of lipogenic regulator genes, fatty acid synthesis genes, and β-oxidation genes in the liver. E: GTT in mice fed NC or NC+Lina diet (top, n = 5) and HF or HF+Lina diet (bottom, n = 8) at 16 weeks of age. F: ITT in mice fed the NC or NC+Lina (top, n = 5) and HF or HF+Lina (bottom, n = 8) diet. G: Plasma insulin levels. H: Immunoblots of phosphorylated Tyr1146 insulin receptor β subunit (p-IRβ), IRβ, phosphorylated Ser473 Akt (p-Akt), and Akt in the eWAT and liver of mice. The levels of p-IRβ and p-Akt were normalized to those of IRβ and Akt, respectively. n = 5–8. *P < 0.05, **P < 0.01 vs. NC group; †P < 0.05, ††P < 0.01 vs. HF group.
Linagliptin alleviates diet-induced insulin resistance and hepatic steatosis. A: Weight gain and tissue weights of C57BL/6J mice fed an NC or HF diet without or with 0.003% linagliptin (Lina) from age 8 to 16 weeks. B: H&E-stained liver sections from mice. Scale bars, 100 μm. C: Liver TG content. D: mRNA expression of lipogenic regulator genes, fatty acid synthesis genes, and β-oxidation genes in the liver. E: GTT in mice fed NC or NC+Lina diet (top, n = 5) and HF or HF+Lina diet (bottom, n = 8) at 16 weeks of age. F: ITT in mice fed the NC or NC+Lina (top, n = 5) and HF or HF+Lina (bottom, n = 8) diet. G: Plasma insulin levels. H: Immunoblots of phosphorylated Tyr1146 insulin receptor β subunit (p-IRβ), IRβ, phosphorylated Ser473 Akt (p-Akt), and Akt in the eWAT and liver of mice. The levels of p-IRβ and p-Akt were normalized to those of IRβ and Akt, respectively. n = 5–8. *P < 0.05, **P < 0.01 vs. NC group; †P < 0.05, ††P < 0.01 vs. HF group.
GTT indicated that linagliptin had no effect on glucose tolerance in NC-fed mice. However, HF diet–induced glucose intolerance was suppressed significantly by linagliptin (Fig. 3E). ITT also showed that HF+Lina mice had increased insulin sensitivity when compared with HF mice (Fig. 3F). Linagliptin also suppressed hyperinsulinemia in both fasting and fed states (Fig. 3G) as well as enhancing insulin signaling in the eWAT and liver of HF-fed mice (Fig. 3H). Thus, linagliptin protected mice against diet-induced hepatic steatosis, insulin resistance, and glucose intolerance.
Linagliptin Attenuates Inflammation in Fat and Liver of DIO Mice
We next investigated the effect of linagliptin on adipose tissue inflammation. Infiltration of macrophages into hypertrophied adipose tissue and crown-like structures that were induced by HF diet decreased markedly in HF+Lina mice according to immunostaining and mRNA expression of F4/80 (Fig. 4A and B). Inflammatory cytokines derived from M1 macrophages, including TNFα, IL-6, and IL-1β, were decreased in eWAT of HF+Lina mice compared with HF mice (Fig. 4B). In addition, the aberrant expression of adipokines such as adiponectin and leptin were improved by linagliptin (Fig. 4B). These findings were associated with attenuated nuclear factor-κB (NF-κB) p65, p38 mitogen-activated protein kinase (MAPK), Jun NH2-terminal kinase (JNK), and extracellular signal–related kinase (ERK) phosphorylation in eWAT of DIO mice (Fig. 4C and Supplementary Fig. 4A). Linagliptin also markedly decreased the number of F4/80+ cells (Fig. 4D and E) and reduced gene expression of proinflammatory cytokines and inflammatory signaling in liver (Fig. 4E and F and Supplementary Fig. 4B).
Linagliptin (Lina) attenuates adipose tissue and liver inflammation in DIO mice. A: Macrophage infiltration in the eWAT of DIO mice, assessed by F4/80 immunostaining. Scale bars, 100 μm. B: mRNA expression of F4/80 and inflammatory cytokines and adipokines in eWAT. C: Immunoblots of phosphorylated p38 MAPK (p-p38 MAPK), phosphorylated NF-κB p65 (p–NF-κB p65), phosphorylated JNK (p-JNK), phosphorylated ERK (p-ERK), and their total proteins in eWAT of mice. D: F4/80 immunostaining in the liver of mice. Scale bars, 100 μm. E: mRNA expression of F4/80 and inflammatory cytokines in the liver. F: Immunoblot of p-p38 MAPK, p–NF-κB p65, p-JNK, p-ERK, and their total proteins in the liver of mice. n = 5–8. *P < 0.05, **P < 0.01 vs. NC group; †P < 0.05, ††P < 0.01 vs. HF group.
Linagliptin (Lina) attenuates adipose tissue and liver inflammation in DIO mice. A: Macrophage infiltration in the eWAT of DIO mice, assessed by F4/80 immunostaining. Scale bars, 100 μm. B: mRNA expression of F4/80 and inflammatory cytokines and adipokines in eWAT. C: Immunoblots of phosphorylated p38 MAPK (p-p38 MAPK), phosphorylated NF-κB p65 (p–NF-κB p65), phosphorylated JNK (p-JNK), phosphorylated ERK (p-ERK), and their total proteins in eWAT of mice. D: F4/80 immunostaining in the liver of mice. Scale bars, 100 μm. E: mRNA expression of F4/80 and inflammatory cytokines in the liver. F: Immunoblot of p-p38 MAPK, p–NF-κB p65, p-JNK, p-ERK, and their total proteins in the liver of mice. n = 5–8. *P < 0.05, **P < 0.01 vs. NC group; †P < 0.05, ††P < 0.01 vs. HF group.
Effects of Linagliptin and Sitagliptin on DPP-4 Activity and Oxidative Stress
Studies have shown that linagliptin at doses of 3–5 mg/kg and sitagliptin at doses of 10–50 mg/kg produce equivalent reductions in plasma glucose, with linagliptin eliciting a greater effect on vascular dysfunction in rodents (18,19). Therefore, the effects of linagliptin and sitagliptin were investigated in mice. Liver and eWAT weights were unaffected by linagliptin or sitagliptin in HF-fed mice (Table 1 and Fig. 5A). Plasma lipid, AST, and ALT were decreased with linagliptin and similarly with sitagliptin (Table 1). Linagliptin produced a marked reduction in hepatic steatosis and TG accumulation in DIO mice, whereas there were only slight reductions following sitagliptin treatment (Fig. 5B). The GTT and ITT data showed that linagliptin and sitagliptin lowered blood glucose by similar amounts (Fig. 5C and D). Hyperinsulinemia and homeostasis model assessment of insulin resistance were also suppressed considerably by both agents (Fig. 5E). Plasma active GLP-1 levels were increased by both linagliptin and sitagliptin treatment (Fig. 5E).
Effects of DPP-4 inhibitors on metabolic parameters at 14 weeks of treatment
. | NC . | HF . | HF+Lina . | HF+Sita . |
---|---|---|---|---|
Body weight (g) | 27.6 ± 0.7 | 43.9 ± 1.3** | 39.3 ± 2.2** | 41.7 ± 2.0** |
Food intake (kcal/day/kg body weight) | 430.2 ± 3.7 | 443.5 ± 43.3 | 489.8 ± 16.0 | 443.3 ± 52.9 |
Liver weight (g) | 1.11 ± 0.03 | 1.41 ± 0.10* | 1.18 ± 0.06 | 1.34 ± 0.17 |
eWAT weight (g) | 0.47 ± 0.07 | 2.24 ± 0.17** | 2.06 ± 0.14** | 2.14 ± 0.16** |
Plasma TG (mg/dL) | 26.3 ± 2.2 | 45.9 ± 5.3* | 35.7 ± 1.2**† | 38.6 ± 1.4** |
Plasma TC (mg/dL) | 72.6 ± 2.8 | 144.8 ± 7.6** | 104.9 ± 11.9*† | 114.8 ± 9.8**† |
Plasma NEFA (mEq/L) | 0.16 ± 0.02 | 0.41 ± 0.02** | 0.34 ± 0.02**† | 0.36 ± 0.02** |
Plasma AST (IU/L) | 14.5 ± 0.6 | 19.1 ± 0.4** | 15.2 ± 1.0†† | 16.8 ± 1.08 |
Plasma ALT (IU/L) | 5.24 ± 0.19 | 6.96 ± 0.48* | 4.25 ± 0.33†† | 5.15 ± 0.47† |
. | NC . | HF . | HF+Lina . | HF+Sita . |
---|---|---|---|---|
Body weight (g) | 27.6 ± 0.7 | 43.9 ± 1.3** | 39.3 ± 2.2** | 41.7 ± 2.0** |
Food intake (kcal/day/kg body weight) | 430.2 ± 3.7 | 443.5 ± 43.3 | 489.8 ± 16.0 | 443.3 ± 52.9 |
Liver weight (g) | 1.11 ± 0.03 | 1.41 ± 0.10* | 1.18 ± 0.06 | 1.34 ± 0.17 |
eWAT weight (g) | 0.47 ± 0.07 | 2.24 ± 0.17** | 2.06 ± 0.14** | 2.14 ± 0.16** |
Plasma TG (mg/dL) | 26.3 ± 2.2 | 45.9 ± 5.3* | 35.7 ± 1.2**† | 38.6 ± 1.4** |
Plasma TC (mg/dL) | 72.6 ± 2.8 | 144.8 ± 7.6** | 104.9 ± 11.9*† | 114.8 ± 9.8**† |
Plasma NEFA (mEq/L) | 0.16 ± 0.02 | 0.41 ± 0.02** | 0.34 ± 0.02**† | 0.36 ± 0.02** |
Plasma AST (IU/L) | 14.5 ± 0.6 | 19.1 ± 0.4** | 15.2 ± 1.0†† | 16.8 ± 1.08 |
Plasma ALT (IU/L) | 5.24 ± 0.19 | 6.96 ± 0.48* | 4.25 ± 0.33†† | 5.15 ± 0.47† |
Data were obtained from 22-week-old mice on different diets. Data are presented as mean ± SEM. n = 6–8.
*P < 0.05, **P < 0.01 vs. mice fed the NC diet;
†P < 0.05, ††P < 0.01 vs. mice fed the HF diet.
Effects of linagliptin and sitagliptin on insulin resistance, hepatic steatosis, and DPP-4 activity in DIO mice. A: Weight gain of mice from age 8 to 22 weeks. B: H&E-stained liver sections and hepatic TG content. Scale bars: 100 μm. C: GTT. D: ITT. E: Plasma insulin levels, HOMA of insulin resistance (HOMA-IR), and active GLP-1 levels. F: DPP-4 activity levels in plasma, eWAT, and liver. n = 6–8. *P < 0.05, **P < 0.01 vs. NC group; †P < 0.05, ††P < 0.01 vs. HF group; ‡P < 0.05, ‡‡P < 0.01 vs. HF+Lina group. G: DPP-4 activity in plasma (before, 2 h after, or 24 h after last gavage), eWAT, and liver from mice gavaged with vehicle, linagliptin (Lina), or sitagliptin (Sita). n = 6. *P < 0.05, **P < 0.01 vs. vehicle; ††P < 0.01 vs. linagliptin-treated group. H: Urine 8-OHdG levels and TBARS content in eWAT and liver. n = 6–8. *P < 0.05, **P < 0.01 vs. NC group; †P < 0.05, ††P < 0.01 vs. HF group; ‡P < 0.05 vs. HF+Lina group.
Effects of linagliptin and sitagliptin on insulin resistance, hepatic steatosis, and DPP-4 activity in DIO mice. A: Weight gain of mice from age 8 to 22 weeks. B: H&E-stained liver sections and hepatic TG content. Scale bars: 100 μm. C: GTT. D: ITT. E: Plasma insulin levels, HOMA of insulin resistance (HOMA-IR), and active GLP-1 levels. F: DPP-4 activity levels in plasma, eWAT, and liver. n = 6–8. *P < 0.05, **P < 0.01 vs. NC group; †P < 0.05, ††P < 0.01 vs. HF group; ‡P < 0.05, ‡‡P < 0.01 vs. HF+Lina group. G: DPP-4 activity in plasma (before, 2 h after, or 24 h after last gavage), eWAT, and liver from mice gavaged with vehicle, linagliptin (Lina), or sitagliptin (Sita). n = 6. *P < 0.05, **P < 0.01 vs. vehicle; ††P < 0.01 vs. linagliptin-treated group. H: Urine 8-OHdG levels and TBARS content in eWAT and liver. n = 6–8. *P < 0.05, **P < 0.01 vs. NC group; †P < 0.05, ††P < 0.01 vs. HF group; ‡P < 0.05 vs. HF+Lina group.
Although DPP-4 activities in plasma and eWAT were decreased by 94% and 92%, respectively, following linagliptin treatment and by 13% and 69%, respectively, following sitagliptin treatment (Fig. 5F), hepatic DPP-4 activity was inhibited only by linagliptin. To better understand the differences between linagliptin and sitagliptin on DPP-4 inhibition, we conducted a short-term study (Fig. 5G). After 5 days’ administration of each DPP-4 inhibitor, plasma DPP-4 activity was almost completely eliminated 2 h after the last treatment with linagliptin and decreased by 68% by sitagliptin. Furthermore, significant reduction of plasma DPP-4 activity was observed only in the linagliptin group 24 h after the last treatment. Similarly, linagliptin showed significant reduction of DPP-4 activities in both eWAT and liver (Fig. 5G).
In parallel to the activity of DPP-4, 8-OHdG, a marker of oxidized DNA damage, was significantly increased in the urine of HF-fed mice, but linagliptin and sitagliptin reduced these levels by 49% and 34%, respectively (Fig. 5H). In addition, increased lipid peroxidation, assessed by TBARS in eWAT and liver, was suppressed by linagliptin but unaffected by sitagliptin (Fig. 5H). These findings occurred in association with increased mRNA expression of mitochondrial fatty acid β-oxidation genes (PPARα, Cpt1α, LCAD, and Acox1) and decreased mRNA expression of NADPH oxidase subunits (gp91phox, p22phox, p67phox, and p47phox) in eWAT and liver of DIO mice (Supplementary Fig. 5).
DPP-4 Inhibition Causes Reciprocal Decrease in M1 Macrophages and Increase in M2 Macrophages in Fat and Liver of Mice
The effect of DPP-4 inhibition on macrophage polarization in vivo was examined. The increased total numbers of ATMs associated with HF feeding were not significantly affected by either linagliptin or sitagliptin treatment (Fig. 6A and B). HF-fed mice showed a significantly higher M1 and lower M2 expression within ATMs. Linagliptin administration resulted in a 28% decrease in M1 ATMs and a 33% increase in M2 ATMs compared with HF feeding, which resulted in macrophage polarization toward an anti-inflammatory phenotype. In contrast, sitagliptin had little effect on ATM phenotype (Fig. 6A and B).
Decreased M1 and increased M2 macrophages in eWAT and liver of DIO mice due to linagliptin administration. A and B: FACS analysis of the stromal vascular cells of epididymal fat pads of mice fed the NC, HF, HF+Lina, or HF+Sita diet for 14 weeks. A: Representative plot of total macrophages (top) and expression of M1 and M2 macrophages (bottom) of mice. B: Quantification of ATMs, M1 ATMs, and M2 ATMs. Data are total ATM counts, percentages of M1 ATMs, percentages of M2 ATMs, and M1/M2 ratios. C and D: FACS analysis of the hepatic nonparenchymal cell fractions. C: Representative plot of total macrophages (top) and expression of M1 and M2 LMs (bottom). D: Quantification of total macrophage counts, percentages of M1 LMs, percentages of M2 LMs, and M1/M2 ratios. n = 6–8. *P < 0.05, **P < 0.01 vs. NC group; †P < 0.05, ††P < 0.01 vs. HF group; ‡P < 0.05 vs. HF+Lina group.
Decreased M1 and increased M2 macrophages in eWAT and liver of DIO mice due to linagliptin administration. A and B: FACS analysis of the stromal vascular cells of epididymal fat pads of mice fed the NC, HF, HF+Lina, or HF+Sita diet for 14 weeks. A: Representative plot of total macrophages (top) and expression of M1 and M2 macrophages (bottom) of mice. B: Quantification of ATMs, M1 ATMs, and M2 ATMs. Data are total ATM counts, percentages of M1 ATMs, percentages of M2 ATMs, and M1/M2 ratios. C and D: FACS analysis of the hepatic nonparenchymal cell fractions. C: Representative plot of total macrophages (top) and expression of M1 and M2 LMs (bottom). D: Quantification of total macrophage counts, percentages of M1 LMs, percentages of M2 LMs, and M1/M2 ratios. n = 6–8. *P < 0.05, **P < 0.01 vs. NC group; †P < 0.05, ††P < 0.01 vs. HF group; ‡P < 0.05 vs. HF+Lina group.
The total number of liver macrophages (LMs) also increased in mice fed the HF diet compared with NC-fed mice (Fig. 6C and D). In addition to reduced total LM content, both linagliptin- and sitagliptin-treated mice had fewer M1 LMs and more M2 LMs than HF-fed mice, which resulted in macrophage polarization toward an anti-inflammatory phenotype in the liver (Fig. 6C and D). The total numbers of CD3+, CD4+, and CD8+ T cells in eWAT and liver increased with HF feeding, and this effect was significantly decreased by linagliptin, whereas it was slightly decreased by sitagliptin (Supplementary Fig. 6). In contrast, there was a significant increase in the Ly6Chi monocyte population in the blood of HF-fed mice relative that of NC-fed mice (Supplementary Fig. 7A), indicating that there was an enhanced recruitment of inflammatory monocytes into the eWAT and liver following HF feeding. However, a predominance of the Ly6C− over the Ly6Chi monocyte population was not observed in either the peripheral blood or the bone marrow of linagliptin- or sitagliptin-treated mice (Supplementary Fig. 7).
Loss of Protective Effects of DPP-4 Inhibitor on Obesity-Related Inflammation and Insulin Resistance in Chemokine-Deficient Mice
Much like incretin peptides, chemokines also act as DPP-4 substrates (10). Among the chemokines, macrophage inflammatory protein (MIP)-1α becomes the most efficient monocyte chemoattractant after cleavage by DPP-4 (25,26), which indicates that DPP-4 may regulate inflammation by increasing the activity of MIP-1α. In the current study, HF diet–induced glucose intolerance, insulin resistance, and hyperinsulinemia were significantly improved in MIP-1α−/− mice compared with wild-type mice (Supplementary Fig. 8A–C). Moreover, macrophage accumulation in the eWAT and liver were decreased by an MIP-1α deficiency with polarization toward an anti-inflammatory phenotype (Supplementary Fig. 8D and E). To further investigate whether the linagliptin-mediated amelioration of macrophage polarization and insulin resistance depended on MIP-1α, MIP-1α−/− mice were fed an HF diet either with or without linagliptin for 8 weeks. Linagliptin did not affect body weight (Fig. 7A) but significantly reduced DPP-4 activities in the plasma, eWAT, and liver by 92%, 84%, and 78%, respectively (Fig. 7B); it also resulted in levels close to those seen in linagliptin-treated wild-type mice. However, linagliptin did not have any significant effects on glucose tolerance, insulin sensitivity, or plasma insulin levels in obese MIP-1α−/− mice (Fig. 7C–E). The total number of ATMs and the percentages of M1 and M2 macrophages in the eWAT and liver were unaffected by linagliptin, which was in accordance with the pattern of F4/80 immunostaining (Fig. 7F and G). The effects of DPP-4 and linagliptin on the regulation of M1/M2 polarization were not observed in peritoneal macrophages obtained from MIP-1α−/− mice (Supplementary Fig. 9).
Loss of protection by linagliptin against obesity-related inflammation and insulin resistance in MIP-1α−/− mice. A: Weight gain of MIP-1α−/− mice on an HF diet with or without linagliptin. B: DPP-4 activities in plasma, eWAT, and liver. C: GTT. D: ITT. E: Plasma insulin levels. F and G: F4/80 immunostaining (left) and quantification of total macrophage counts and percentages of M1 macrophages and M2 macrophages (right) in eWAT (F) and liver (G) by FACS. n = 8. Scale bars: 100 μm. **P < 0.01 vs. HF group.
Loss of protection by linagliptin against obesity-related inflammation and insulin resistance in MIP-1α−/− mice. A: Weight gain of MIP-1α−/− mice on an HF diet with or without linagliptin. B: DPP-4 activities in plasma, eWAT, and liver. C: GTT. D: ITT. E: Plasma insulin levels. F and G: F4/80 immunostaining (left) and quantification of total macrophage counts and percentages of M1 macrophages and M2 macrophages (right) in eWAT (F) and liver (G) by FACS. n = 8. Scale bars: 100 μm. **P < 0.01 vs. HF group.
Discussion
Our study provides firm evidence for DPP-4 playing a crucial role in regulating the macrophage-mediated inflammatory response to obesity and the development of insulin resistance. We demonstrated how DPP-4 is predominantly expressed in macrophages, particularly M1 macrophages, rather than in adipocytes in WAT. We also found that administration of the DPP-4 inhibitor linagliptin attenuates oxidative stress, inflammation, and insulin resistance, at least in part, through reduction of macrophage accumulation and alternative macrophage activation in both WAT and liver in DIO mice. Moreover, despite equivalent reductions in blood glucose, we showed that linagliptin reduces markers of oxidative stress to a greater extent than sitagliptin. Interestingly, the apparently protective effects of linagliptin are abrogated in MIP-1α–deficient mice, suggesting that linagliptin may alleviate obesity-associated inflammation partly due to its actions on a chemokine MIP-1α–dependent mechanism.
Recent studies show that DPP-4 expression and release from visceral fat is augmented in obese and insulin-resistant subjects (12–14). Furthermore, plasma DPP-4 activity is positively correlated with HbA1c levels in patients with type 2 diabetes (27,28). However, the cellular sources responsible for DPP-4 in WAT remain unclear. Our findings suggest that dominant DPP-4 expression in SVF of WAT (Fig. 1A and B) may derive from several types of stromal vascular cells, including preadipocytes, infiltrated macrophages, and other hematopoietic cells. However, consistent with previous findings (14), our results show that DPP-4 is expressed mainly on F4/80+ macrophages rather than adipocytes (Fig. 1B and C). FACS analysis also revealed that DPP-4+ ATMs increase in obese mice compared with lean mice (Fig. 1D). Similar to a previous report (14), high levels of DPP-4 were detected on adipose macrophages (Fig. 1C and D) but not on T cells (data not shown) in obese mice. This observation suggests that DPP-4+ macrophages are a major source of circulating DPP-4 in the obese state.
Dysregulation of M1/M2 polarization in macrophages is emerging as a central mechanism underlying the pathogenesis of obesity and comorbidities such as insulin resistance and nonalcoholic fatty liver disease (29,30). Deletion of M1 macrophages normalizes sensitivity to insulin in obese mice (3,31), whereas reducing the number of M2 macrophages predisposes lean mice to insulin resistance (4). M2-type Kupffer cells in the liver serve to protect against nonalcoholic fatty liver disease (32). Thus, strategies restraining M1 polarization and/or driving alternative M2 activation of macrophages may have the potential to protect against exacerbated inflammation and insulin resistance and attenuate the progression to steatohepatitis. Our in vitro findings (Fig. 2) and FACS data (Fig. 6) indicate that DPP-4 inhibition causes an anti-inflammatory macrophage polarization in ATMs and LMs, which contributes to the attenuation of whole-body insulin resistance. Furthermore, DPP-4 per se induced M1 polarization but suppressed M2 polarization of macrophages (Fig. 2A and C), whereas the dysregulation of M1/M2 status was reversed by linagliptin (Fig. 2B and D). These data imply that there is a direct link between DPP-4 activation and M1 polarization in the macrophage. Elevated levels of plasma GLP-1 could represent the mechanism underlying this process, and this was reflected in the inhibition of DPP-4 that accompanied chronic inflammation. However, linagliptin increased plasma GLP-1 in DIO mice (Fig. 5E), but only to a small extent compared with exogenous GLP-1 administration (10). Several lines of evidence suggest that the consequences of DPP-4 inhibition are far more complex than previously thought and may involve both GLP-1–dependent and –independent effects (8,33–35). Considering its pattern of expression and the multiplicity of functions and targets of DPP-4, DPP-4 may play a distinct role in regulating macrophage polarity in addition to its effect on the incretin axis.
The infiltration of Th1 and CD8+ T cells precedes M1-polarized macrophage recruitment, and interactions between T cells and macrophages constitute a maladaptive feed-forward loop, which leads to adipose inflammation and insulin resistance (36,37). Therefore, DPP-4 inhibition may reduce the accumulation of T cells as well as M1 activation of macrophage to alleviate insulin resistance and inflammation in obesity (Supplementary Fig. 6). In humans with type 2 diabetes, treatment with DPP-4 inhibitors decreases proinflammatory markers in the blood and inhibits NF-κB activation in mononuclear cells (38). However, DPP-4 inhibitor treatment might be a double-edged sword (39), as it may increase the risk of infection in patients with diabetes (40). Currently, the long-term effects of DPP-4 inhibitors on T cell maturation and activation remain unknown, and further research is needed to clarify their long-term immunological effects.
An important question is whether DPP-4 inhibitors can regulate the recruitment of monocytes and affect M1/M2 status in bone marrow or peripheral blood given the link between Ly6Chi/Ly6C− monocyte subtypes and their fate as M1/M2 macrophages (41–43). In the current study, HF feeding caused a significant increase in recruited inflammatory Ly6Chi monocytes in the peripheral blood, whereas DPP-4 inhibitors did not affect Ly6Chi or Ly6C− monocytes either in bone marrow or peripheral blood (Supplementary Fig. 7). According to the FACS analysis, although the number of crown-like structures revealed by F4/80 immunostaining was decreased by linagliptin, the number of total ATMs was unaffected by DPP-4 inhibitors. This inconsistency can be explained by differences in the antibodies used for the immunostaining and flow cytometry procedures. Thus, inhibiting DPP-4 caused a dynamic M2 shift of macrophages within WAT and liver, which contributed to the attenuation of local and systemic insulin resistance. In contrast, LMs, which consist of resident Kupffer cells and recruited bone marrow–derived macrophages, decreased in HF-fed mice following treatment with linagliptin (Fig. 6C and D), which suggests that the reduction of LMs was mainly due to the decreased activation of resident Kupffer cells.
Increased oxidative stress causes MCP-1 production from accumulated fat, which, in obese subjects, leads to infiltration of ATMs (2,44). A major contributor to oxidative stress in fat and vasculature (44,45), NADPH oxidase can activate both NF-κB and MAPK subfamilies, thus increasing inflammatory response. In this study, linagliptin decreased ROS generation and suppressed the expression of NADPH oxidase subunits in both WAT and liver (Fig. 5H and Supplementary Fig. 5). Furthermore, DPP-4 activity, ROS production, and M1/M2 polarity were associated (Fig. 2 and Supplementary Fig. 2). Thus, the anti-inflammatory effects of DPP-4 inhibition are due to the attenuation of oxidative stress from accumulated fat or infiltrated macrophages in WAT and/or liver.
It is possible that the anti-inflammatory effects of linagliptin are attributable to its impact on glucose control mediated by DPP-4 inhibition rather than any other molecule-specific properties (46,47). When comparing linagliptin and sitagliptin, although we find that both DPP-4 inhibitors lower glucose levels to an equivalent extent in DIO mice, linagliptin elicited a greater DPP-4 inhibition and may also have had stronger antioxidative (Fig. 5) and anti-inflammatory actions (Fig. 6). The diversity of the drug effects may result from differences in their tissue availability and/or inhibitory efficiencies against DPP-4. Linagliptin is highly tissue penetrative, whereas sitagliptin is only weakly tissue penetrative (47,48). Linagliptin also has a longer duration of DPP-4 inhibition, something we have shown in our own work (Fig. 5G) and that has been observed previously (46).
DPP-4 is a 766-amino acid serine protease that preferentially cleaves N-terminal dipeptides from various substrates (26). Among them, MIP-1α, also known as CCL3, is converted into an efficient monocyte/macrophage attractant after cleavage by DPP-4 (25,26). MIP-1α is robustly upregulated in WAT of obese mice, and infiltrated ATMs or preadipocytes can secrete MIP-1α in inflamed fat (49). The present results showed that the loss of MIP-1α resulted in improved glucose homeostasis and the reduction of macrophage accumulation as well as in a polarization toward an anti-inflammatory phenotype in the WAT and liver (Supplementary Fig. 8). Linagliptin did not confer the protection against obesity-induced inflammation or insulin resistance in MIP-1α−/− mice despite DPP-4 activity being markedly inhibited by linagliptin (Fig. 7). Furthermore, DPP-4 unaffected M1/M2 marker mRNA expression or ROS production in peritoneal macrophages from MIP-1α−/− mice (Supplementary Fig. 9), suggesting that MIP-1α may be a substrate for DPP-4 that, at least in part, contributes to the regulation of macrophage polarization under conditions of obesity.
In conclusion, our findings suggest that DPP-4 plays a critical role in obesity-induced inflammation and insulin resistance by regulating the M1/M2 status of macrophages. DPP-4+ macrophages accumulate in WAT of obese mice, and, importantly, inhibition of DPP-4 with linagliptin results in macrophage polarization toward an anti-inflammatory phenotype in adipose tissue and liver, thereby attenuating obesity-induced inflammation and insulin resistance. Overall, the current investigation highlights a potential clinical utility for DPP-4 inhibition in the attenuation of macrophage-mediated inflammation and prevention of insulin resistance and type 2 diabetes.
Article Information
Acknowledgments. The authors thank M. Nakayama and K. Hara (Kanazawa University, Kanazawa, Japan) for technical assistance and animal care and Tim Hardman of Niche Science & Technology for help in the preparation of the manuscript.
Funding. This work was supported by Grant-in-Aid for Scientific Research (B) (25282017) and Challenging Exploratory Research (15K12698) from the Ministry of Education, Culture, Sports, Science, and Technology of Japan and Boehringer Ingelheim Pharma GmbH & Co. KG grant (to T.O.).
Duality of Interest. T.O. received research support from Boehringer Ingelheim Pharma GmbH & Co. KG. No other potential conflicts of interest relevant to this article were reported.
Author Contributions. F.Z., Y.N., M.N., N.N., and L.X. performed experiments and acquired data. N.M. and S.K. contributed to discussion and edited the manuscript. T.O. contributed to the study concept and design and wrote the manuscript. T.O. is the guarantor of this work and, as such, had full access to all 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 75th Scientific Sessions of the American Diabetes Association, Boston, MA, 5–9 June 2015.