|Home | About | Journals | Submit | Contact Us | Français|
Metformin has been reported to increase the expression of the glucagon-like peptide-1 (GLP-1) receptor in pancreatic beta cells in a peroxisome proliferator-activated receptor (PPAR)-α-dependent manner. We investigated whether a PPARα agonist, fenofibrate, exhibits an additive or synergistic effect on glucose metabolism, independent of its lipid-lowering effect, when added to metformin. Non-obese diabetic Goto-Kakizaki (GK) rats were divided into four groups and treated for 28 days with metformin, fenofibrate, metformin plus fenofibrate or vehicle. The random blood glucose levels, body weights, food intake and serum lipid profiles were not significantly different among the groups. After 4 weeks, metformin, but not fenofibrate, markedly reduced the blood glucose levels during oral glucose tolerance tests, and this effect was attenuated by adding fenofibrate. Metformin increased the expression of the GLP-1 receptor in pancreatic islets, whereas fenofibrate did not. During the intraperitoneal glucose tolerance tests with the injection of a GLP-1 analog, metformin and/or fenofibrate did not alter the insulin secretory responses. In conclusion, fenofibrate did not confer any beneficial effect on glucose homeostasis but reduced metformin's glucose-lowering activity in GK rats, thus discouraging the addition of fenofibrate to metformin to improve glycemic control.
Relative insulin deficiency in the presence of prevailing insulin resistance is the main pathophysiology of type 2 diabetes mellitus.1 The mechanisms of insulin secretion in response to a mixed meal are complex.2 Among these mechanisms, the incretin effect contributes 50–70% of the insulin secretion in subjects with normal glucose tolerance.3, 4 The incretin hormones, glucagon-like peptide-1 (GLP-1) and glucose-dependent insulinotropic polypeptide (GIP) are secreted from intestinal endocrine L- and K-cells, respectively, following the ingestion of nutrients and enhance insulin secretion from pancreatic beta cells in a glucose-dependent manner.5, 6, 7 The effects of GLP-1 and GIP terminate shortly after secretion due to the enzymatic activity of dipeptidyl peptidase-4.6 Intriguingly, the incretin effect is markedly attenuated by as much as 10–40% in patients with type 2 diabetes mellitus,8, 9, 10 which can be explained not only by decreased beta cell function/mass per se but also by decreased incretin sensitivity due to the downregulation of the relevant receptors.11, 12, 13 Therefore, it may be possible to control the glucose level in type 2 diabetes mellitus patients by increasing the incretin sensitivity by upregulating the expression of incretin receptors in pancreatic beta cells.14
Incretin therapies such as GLP-1 analogs and dipeptidyl peptidase-4 inhibitors are currently available in clinical practice.15, 16 This type of new treatment has the advantages of low risks of hypoglycemia and weight gain relative to current anti-diabetes treatments based on sulfonylureas or insulin.17 It is conceivable that the glucose-lowering effects of such incretin therapies would be augmented if they were combined with medications that effectively increase incretin sensitivity through the upregulation of incretin receptor expression. In addition to increased insulin secretion, we may expect an enhancement of other beneficial effects of incretin hormones, such as the preservation of or an increase in beta cell mass or function, delayed gastric emptying and decreased appetite.6, 17
The addition of GLP-1 analogs or dipeptidyl peptidase-4 inhibitors to metformin provides further glucose-lowering effects without causing hypoglycemia or weight gain.18, 19, 20 The glucose-lowering effect of metformin is based on the suppression of hepatic glucose production and the enhancement of peripheral glucose utilization.14, 21 GLP-1 increases insulin secretion from pancreatic beta cells and decreases glucagon secretion from pancreatic alpha cells, which in turn leads to decreased hepatic glucose output.22 Therefore, both metformin and incretin therapy converge to regulate hepatic glucose output. In addition, Maida et al.23 recently demonstrated that metformin increases GLP-1 secretion from intestinal endocrine L cells and also increases the expression of the GLP-1 receptor in pancreatic islets via a peroxisome proliferator-activated receptor (PPAR)-α-dependent pathway. Furthermore, in an obese type 2 diabetes animal model, fenofibrate (a PPARα agonist) reduced the decline in beta cell mass and prevented the development of diabetes.24 Based on these results, we hypothesized that a PPARα agonist, fenofibrate, may enhance the effect of GLP-1 when used either as a monotherapy or in combination with metformin.14 To explore this possibility, we examined the effect of fenofibrate alone or in combination with metformin on glucose homeostasis and GLP-1 sensitivity in non-obese Goto-Kakizaki (GK) rats.
The animal studies were performed after receiving approval from the Institutional Animal Care and Use Committee (IACUC) of Seoul National University (IACUC approval no. 11−0038). Twenty-four male GK rats (aged 5 weeks) were obtained from Japan SLC (Hamamatsu, Japan). They were provided with standard chow (Purina rat and mouse chow, Purina Korea, Seoul, Korea) and water ad libitum at the Seoul National University Hospital Biomedical Research Institute. After a 1-week adaptation period, the GK rats were divided into four groups (n=6 in each group): the vehicle-treated control (Ctrl) group, the metformin alone (MA) group, the fenofibrate alone (FA) group and metformin plus fenofibrate (MF) group. The vehicle for both metformin and fenofibrate was 1% carboxymethyl cellulose (Sigma-Aldrich, St Louis, MO, USA). Each medication was administered by gastric gavage every day for 28 days. Metformin (Glucophage; Merck Serono, Geneva, Switzerland) was administered at a dose of 150mgkg−1 for 14 days, and then the dose was increased to 300mgkg−1 during the last 14 days. The dose of fenofibrate (Recipharm Fontaine, Dijon, France) was fixed at 150mgkg−1 for 28 days.
The body weights and random blood glucose levels were monitored every day, and food intake over 24h was measured every week. The body weights and food intake were measured using an electrical scale (A&D, Tokyo, Japan), and the blood glucose levels were measured using a glucometer (OneTouch Ultra; LifeScan, Milpitas, CA, USA) and tail vein blood samples. The total blood cholesterol and triglyceride levels were measured enzymatically with an autoanalyzer (Hitachi 7070; Hitachi, Tokyo, Japan).
Oral glucose tolerance tests (OGTTs) were performed after 12-h fasts following 4 weeks of treatment. The baseline blood glucose levels were measured, and each medication was administered by gastric gavage 1h before oral glucose loading. The OGTT consisted of 2.0gkg−1 of glucose in a 20% solution delivered by gastric gavage. The glucose levels were monitored at 0, 15, 30, 60, 120 and 180min using a glucometer (OneTouch Ultra) and tail vein blood samples.
To evaluate the insulin secretory capacity in response to exendin-4, a GLP-1 analog, IPGTTs with exendin-4 injections were performed after 12-h fasts 2 days after the OGTTs. The baseline blood glucose levels were measured 1h before glucose loading, and then exendin-4 (Sigma-Aldrich) at a dose of 25nmolkg−1 was injected via the intraperitoneal (i.p.) route. An hour after the i.p. exendin-4 injection, a 2.0gkg−1 load of glucose as a 20% solution was injected in the same way. The study medications were not administered on the day of the IPGTT to avoid the effect of the medications. Blood samples (150μl) for glucose measurements and plasma samples for insulin measurements were obtained at 0, 30, 60, 120 and 180min from the tail vein. The insulin levels were measured using a commercial ELISA kit (Alpco Diagnostics, Windham, NH, USA).
Following completion of the OGTTs and IPGTTs, the GK rats were killed by an i.p. injection of Zoletil 50 (zolazepam plus tiletamine; Virbac Laboratories, Carros, France) and xylazine (Bayer Korea, Seoul, Korea). The pancreas of each animal was fixed with 4% formalin solution and embedded in paraffin. The samples were exposed to an anti-insulin antibody (1:1000, mouse monoclonal antibody, I-2018, Sigma-Aldrich) followed by 5% normal goat serum (S-1000, Vector Laboratories, Burlingame, CA, USA). The sections were incubated with a secondary mouse antibody (1:1000, biotinylated anti-mouse IgG(H+L), BA-9200, Vector Laboratories) for 1h. Then, the samples were reacted with an avidin–biotin complex (Vectastain Elite ABC kit, PK-6100, Vector Laboratories) for 30min and exposed to 3,3′-diaminobenzidine (SK-4100, Vector Laboratories) for 20s. The entire slide was scanned using an Aperio ScanScope (Aperio Technologies, Vista, CA, USA), and the beta cell area was quantified by counting the number of pixels with insulin staining using Aperio ImageScope. To measure the GLP-1 receptor expression in pancreatic islets, we performed immunofluorescent staining for the GLP-1 receptor and insulin. The samples were exposed to a primary GLP-1 receptor antibody (1:100, rabbit polyclonal GLP1R, ab39072, Abcam, Cambridge, UK) and an insulin antibody (1:1000, mouse monoclonal antibody, I-2018) for 12h. Next, the samples were treated with Chromeo 488 (1:500, goat polyclonal secondary antibody to rabbit IgG(H+L), ab60314, Abcam) to visualize the bound GLP-1 receptor antibody and Alexa 594 (1:500, Alexa Fluor 594 goat anti-mouse IgG(H+L), A-11005, Invitrogen) to visualize the bound insulin antibody. Images were captured by a BX-50 Olympus microscope (Olympus, Tokyo, Japan) and quantified using Aperio ImageScope (Aperio Technologies). The level of GLP-1 receptor expression was expressed as the area of GLP-1 receptor staining/the area of insulin staining.
The data analysis was performed using Prism 5.0 (GraphPad, San Diego, CA, USA) and SPSS (Statistical Package for the Social Sciences) version 17.0 software (SPSS, Chicago, IL, USA). The initial blood glucose levels and body weights and the food intake were analyzed by nonparametric analysis using Kruskal–Wallis tests followed by Games–Howell post hoc tests. Two-way repeated measures analysis of variance with the Bonferroni post hoc test was used to analyze time-course differences in the glucose levels and body weights. The AUC of glucose and the incremental AUC (iAUC) of insulin were calculated using the trapezoidal rule. The AUC of glucose, the iAUC of insulin, the insulin-positive area, and the GLP-1 receptor-positive area were analyzed by the nonparametric Kruskal–Wallis test followed by the Games–Howell post hoc test. P-values<0.05 were considered significant.
The baseline body weights were not significantly different among the four groups. The changes in body weight over time also did not differ among the four groups (P=0.128), but the Bonferroni post hoc test revealed differences between the Ctrl and MF groups from day 21 (Figure 1a). Food intake was comparable among the groups (Figure 1b). The random blood glucose levels at baseline and during the 4-week treatment period were similar among the groups (Figure 1c). The total cholesterol and triglyceride levels were comparable among the groups (Figure 1d), which indicated that fenofibrate had no effect on the plasma levels of total cholesterol and triglycerides in non-obese diabetic GK rats.
After 4 weeks of treatment, OGTTs were performed with the administration of the study medications (Figure 2a). The blood glucose levels were lower at 15, 30, 60 and 120min in the MA group than in the Ctrl group. At 120min, the glucose levels were higher in the FA group than in the Ctrl group. The MF group had lower blood glucose levels at 15 and 30min than the Ctrl group. However, the MF group had higher blood glucose levels at 60, 120 and 180min than the MA group. The FA group had higher blood glucose levels than the MA group from 30min to the end of the test. The AUC of the glucose excursions over a 240-min period (from −60min to +180min) was significantly lower in the MA group than in the Ctrl group, whereas it was higher in the FA group than in the MA group (Figure 2b).
To examine the GLP-1 sensitivity, we performed IPGTTs with i.p. exendin-4. The glucose levels during the IPGTTs with i.p. exendin-4 injections did not differ between the Ctrl and MA groups but were higher in the FA and MF groups at 120 and 180min. The blood glucose levels at 120 and 180min were also higher in the FA and MF groups than in the MA group (Figure 3a). The AUCs for blood glucose during the IPGTTs were higher for the FA and MF groups than for the Ctrl group, and these values were higher in the MF group than in the MA group (Figure 3b).
The plasma insulin levels during the IPGTTs were lower between 30 and 180min in the MF group and at 180min for both the MA and FA groups than in the Ctrl group (Figure 3c). There were no significant differences in the iAUCs for the plasma insulin levels among the groups (Figure 3d). However, compared with the Ctrl or MA group, the FA and MF groups had lower iAUCs for the plasma insulin levels, but these differences did not reach statistical significance (Figure 3d).
Representative images of pancreatic islets stained for insulin after 4 weeks of treatment are shown in Figure 4a. The morphology of the islets of GK rats were typically distorted, as reported elsewhere.25 The average insulin-positive area corrected by the total pancreatic area was 1.3±0.2% in the Ctrl group, 1.1±0.2% in the MA group, 1.1±0.2% in the FA group and 0.7±0.1% in the MF group; thus, the values for the different groups were comparable. According to the post hoc analysis, the average percentage of the total pancreas area positive for insulin tended to be lower in the MF group than in the Ctrl group (P=0.108) (Figure 4b).
Representative images of pancreatic islets immunostained for insulin (red fluorescence) and the GLP-1 receptor (green fluorescence) are shown in Figure 4c. Most GLP-1 receptor immunoreactivity overlapped with the insulin immunostaining. The average GLP-1 receptor-positive area per insulin-positive area was significantly higher in the MA group than in the Ctrl group (45.4±3.3% vs 26.1±2.1%, P<0.001). The average percentage of GLP-1 receptor-positive area per insulin-positive area in the FA group was lower than that in the MA group (22.6±1.7% vs 45.4±3.3%, P<0.001), and that in the MF group was higher than that in the FA group (34.3±3.7% vs 22.6±1.7%, P=0.029) (Figure 4d).
In this study, we examined whether a PPARα agonist, fenofibrate, with or without metformin can improve glucose homeostasis through enhancing GLP-1 sensitivity by increasing the expression of the GLP-1 receptor in the pancreatic beta cells of non-obese diabetic GK rats. There were no differences in body weight and food intake among the groups during the study period. The total cholesterol and triglyceride levels were comparable among the four treatment groups, which allowed us to examine the effect of fenofibrate on glucose homeostasis independent of its lipid-lowering effects in non-obese diabetic GK rats.
Although metformin lowered the blood glucose levels during the OGTTs, fenofibrate did not exhibit glucose-lowering effects, but rather, caused a higher glucose level during the late postprandial period. Similarly, a recent study revealed that 6 weeks of treatment with bezafibrate, a PPARα agonist, did not affect glucose homeostasis in db/db mice, an obese diabetic rodent model.26 Furthermore, in our study, fenofibrate offset the glucose-lowering effect of metformin during the OGTTs, which indicates that fenofibrate has a negative effect on glucose homeostasis when added to metformin. However, in human studies, even though they were not designed to evaluate the glucose-lowering effect, fenofibrate had no adverse effect on glucose homeostasis when added to metformin in type 2 diabetic patients with mixed dyslipidemia27 or in patients with metabolic syndrome.28 In the Fenofibrate Intervention and Event Lowering in Diabetes (FIELD) trial, more than 40% of the study subjects enrolled in each treatment group (placebo vs fenofibrate) were taking metformin, but there was no meaningful change in the HbA1c levels from the baseline values in either group.29 The mechanism responsible for the discrepancy between our experimental results and the results of previous clinical trials needs to be elucidated.
Maida et al.23 reported that short-term (48h) metformin treatment significantly increased GLP-1 receptor expression in the pancreatic islets of non-diabetic mice relative to vehicle-treated mice. In this study, we found that long-term (4 weeks) treatment with metformin also increased the expression of the GLP-1 receptor in pancreatic islets in non-obese diabetic GK rats, which are characterized by prominent beta cell destruction.30 Furthermore, the expression level of the GLP-1 receptor in the pancreatic islets of GK rats was higher when the rats were treated with metformin and fenofibrate than with fenofibrate alone. Taken together, these results indicate that metformin increases the expression of the GLP-1 receptor in pancreatic islets in both non-diabetic and diabetic animal models regardless of whether fenofibrate is used. However, the increased expression of the GLP-1 receptor was not translated into an increased insulin response to an intraperitoneally administered GLP-1 receptor agonist (exendin-4). Severely compromised beta cell function and/or mass in GK rats might compromise the ability of metformin to enhance insulin secretion in response to exendin-4.
It has been reported that metformin-treated mice exhibit increased expression of the GLP-1 receptor in pancreatic islets independent of the effect of AMP-activated protein kinase.23 Instead, the increased expression of the GLP-1 receptor was attributed to the action of PPARα because no such increase was observed in PPARα-knockout mice in vivo and because metformin increased GLP-1 receptor expression in INS-1 beta cells in vitro in a PPARα-dependent fashion.23 In contrast, in our study, fenofibrate, a potent PPARα agonist, did not have any additive or synergistic effect on the expression of the GLP-1 receptor when used in combination with metformin. The concept of the selective modulation of nuclear receptors by different ligands31 may provide a plausible explanation for this apparent discrepancy.
For PPARγ, another PPAR family member, different ligands induce different receptor conformations and therefore different biological responses, as explained by the selective PPAR modulator (SPPARM) model.32 The SPPARM model could be extended to the regulation of PPARα activity. Gemfibrozil, another PPARα agonist available in clinical practice for the treatment of dyslipidemia, behaves as a partial agonist, whereas fenofibrate behaves as a full agonist, recruiting coactivators to the human apoA-I DR-2 site more effectively than gemfibrozil.33 Accordingly, we may be able to explain the divergent regulation of apoA-I in humans by gemfibrozil or fenofibrate.33 In Maida et al.,23 PPARα was modulated by unidentified endogenous ligands that were presumably induced by metformin treatment, whereas we used a synthetic full agonist of PPARα in this study. We speculate that different PPARα ligands might be responsible for the metformin response at the level of GLP-1 receptor expression in pancreatic islets.
In summary, fenofibrate did not confer any beneficial effect on glucose homeostasis but instead reduced metformin's glucose-lowering activity in GK rats. Because GK rats are non-obese and are characterized by decreased beta cell mass associated with extensive islet fibrosis,30 animal models with obesity or mild beta cell dysfunction may have different responses to metformin and/or fenofibrate treatment, and therefore, further investigation is required.
This study was supported by a Chungram research grant from the Korean Association of Internal Medicine and the Basic Science Research Program through the National Research Foundation of Korea (NRF) that was funded by the Ministry of Education, Science and Technology (2011-0009127). We thank Professor Timothy J Kieffer (University of British Columbia, Canada) for his valuable comments on and discussion regarding this manuscript.
The authors declare no conflict of interest.