Uncoupling Protein 2 Knockout Mice Have Enhanced Insulin Secretory Capacity After a High-Fat Diet

Male and female UCP2 (WT) or (−/−) mice were used in the present study and were bred from lines generated previously, as described by us (8). At 4 months of age, half of the mice were placed on either a control diet (Rodent Diet #8664; Harlan Teklad, Madison, WI) or an HFD (#D12451; Research Diets, New Brunswick, NJ). The fat source was lard and composed 45% of the total calories. The mice remained on the HFD or control diet for 4.5 months.

Blood glucose levels were measured using a Lifescan Elite glucose meter (Lifescan, Toronto, ON, Canada). Plasma insulin levels were determined using a rat insulin enzyme-linked immunosorbent assay kit (Crystal Chem, Chicago, IL). Plasma glucagon levels were measured by radioimmunoassay (Linco, St. Charles, MO). Plasma FFAs were measured using a NEFA C FFA kit (Wako Chemicals USA, Richmond, VA). Plasma triglycerides were measured using a BioScanner 2000 Triglyceride system (Polymer Technologies Systems, Indianapolis, IN). Intraperitoneal glucose tolerance tests (IPGTTs) and insulin tolerance tests (ITTs) were performed as previously described (8). The area under the curve (AUC) analysis represents the AUC independent of basal values.

Pancreatic islets were isolated as previously described (8). RNA was extracted from cultured islets using TRIzol reagent (Gibco BRL) following the manufacturer’s instructions. Pancreatic insulin and glucagon contents were determined as previously described (13,14).

Pancreatic tissues were fixed in Bouin’s fixative (pH 7.2) for 45 min and embedded in paraffin. Sections were prepared at three levels of the pancreas separated by 100 μm. For insulin staining, the primary antibody was guinea pig anti-insulin (1:1,000; RA Pederson, Vancouver, BC, Canada) and the secondary was rabbit anti-guinea pig peroxidase (1:100; Dako, Mississauga, ON). Slides were labeled with Signet USA Level 2 labeling reagent (Ultra Streptavidin-Horseradish Peroxidase Complex) and developed with Nova Red (Vector Labs, Burlington, ON, Canada). For β-cell replication, 100 mg/kg 5-bromo-2′deoxyuridine (BrdU; Sigma) was injected intraperitoneally 6 h before pancreas removal. After staining for insulin, sections were stained for BrdU with a monoclonal BrdU antibody (Clone IU-4; Caltag Labs, Burlingame, CA). The secondary antibody was biotinylated horse anti-mouse linking antibody (Vector Labs). The labeling reagent was Signet Ultra Streptavidin detection system-Alkaline phosphatase (Signet, Denham, MA). The substrate was BCIP/NBT (Dako). The slides were then counterstained with hematoxylin. For β-cell apoptosis, sections were stained using ApopTag Peroxidase In Situ Apoptosis Detection Kit (Intergen, Purchase, NY), with addition of a 15-min microwave step (medium power) before the proteinase K step. Data were expressed as percentage of positive β-cells for either BrdU or apoptosis. At least 5,000 insulin-positive cells were counted for each treatment. Apoptotic nuclei were identified by both presence of brown nuclei and morphological analysis (15). The percentage of β-cell area in the pancreas was calculated by dividing the area of all insulin-positive cells in one section by the total area of this section and then multiplying this ratio by 100.

Mice were fasted overnight (15–18 h) and anesthetized with 80 mg/kg sodium pentobarbital intraperitoneally. The surgical procedure for the perfusion of the pancreas was modified from that described previously (16,17). In brief, PE 50 tubing (Intramedic, Parsippany, NJ) was used to cannulate the dorsal aorta and the hepatic portal vein. The perfusate was a Krebs-Ringer 2% BSA glucose 3% Dextran solution (17). The solution was gassed with 95% O₂/5% CO₂ to achieve a pH of 7.4 and heated to 37°C. The glucose concentration was delivered as a linear gradient between 1.4 and 20 mmol/l over a 55-min period. Samples were stored at −20°C until assayed for insulin, as described (18).

Medium containing palmitic acid (Sigma) was prepared as a stock solution in 5% fatty acid free BSA (Sigma) and diluted to 0.5 mmol/l on the day of the experiment. Control medium was prepared from a stock solution that did not contain palmitic acid. The FFA concentration of the stock was confirmed by analysis with a NEFA C kit. Islets were incubated for 48 h with or without palmitic acid in the presence of 8.3 mmol/l glucose, after which insulin secretion or ATP concentrations were determined. ATP was assayed as previously described (19). Briefly, islets were extracted in groups of 50 islets with 0.1 mol/l NaOH/0.5 mmol/l EDTA. ATP measurements were performed using a bioluminescent assay kit (Sigma). Mitochondrial membrane potential was measured in isolated islets using JC-1 (Molecular Probes, Eugene, OR) (20). Islets from WT or UCP2 (−/−) mice were loaded with 10 μg/ml JC-1 for 20 min at room temperature in a 96-well fluorescent plate reader (excitation 485, emission 590). Results were normalized by DNA content that was estimated with SYBR Green I dye (Sigma) in islets extracted as according to Laird et al. (21).

Probes for UCP2, CPT-1, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were used to quantify the level of gene expression in isolated WT and UCP2 (−/−) mouse islets (Table 1). UCP2- and GAPDH-specific fluorescent probes were obtained from Biosearch Technologies (Novato, CA). UCP2 and GAPDH expression levels were measured simultaneously using a Taqman Universal PCR master mix and an ABI Prism 7900HT Sequence Detection System (Applied Biosystems, Branchburg, NJ). All mRNA levels were quantified using a standard curve of a known amount of mouse cDNA (cDNAs were created from forward and reverse primers; Table 1). All cDNA PCR products that were used in the standard curve were purified using a MinElute PCR Purification Kit (Qiagen, Mississauga, ON, Canada). Quantification of CPT-1 gene expression levels was performed using a SYBR green detection system (SYBR Green PCR Master Mix, Applied Biosystems). The real-time PCR conditions were 2 min at 50°C, 10 min at 95°C, and then 40 cycles for 15 s at 95°C and 1 min at 53°C. Melt curve analysis demonstrated that each primer set described in Table 1 amplified a single predominant product with a distinct melting temperature (T_m), and the size of the PCR products was confirmed by running the PCR products on a 1.2% agarose gel. UCP2 and CPT-1 gene levels were corrected for GAPDH levels.

Statistical significance was assessed by using either Student’s t test or one-way or two-way ANOVA for repeated measures followed by multiple Bonferroni comparisons using a Statistical Analysis System program (SAS). All data are expressed as means ± SE.

All WT and UCP2 (−/−) mice on HFD gained a significant amount of weight during the treatment (Table 2). The net weight gain of UCP2 (−/−) mice was greater than that of WT mice (P < 0.05; Table 2). There were no differences in fasting blood glucose concentrations for control diet WT and UCP2 (−/−) mice (Fig. 1A). HFD WT mice significantly increased their fasting blood glucose compared with control diet WT mice, but this increase did not occur in UCP2 (−/−) mice (Fig. 1A). UCP2 (−/−) mice also had significantly lower fed blood glucose concentrations compared with WT mice fed the control diet (P < 0.01) and fed the HFD (P < 0.001; Fig. 1B). Fasting insulin concentrations in the control diet UCP2 (−/−) mice were significantly (P < 0.01) higher than the control diet WT mice as previously reported (8) (Fig. 1C). HFD fasting plasma insulin concentrations rose significantly in both WT (P < 0.001) and UCP2 (−/−) mice (P < 0.01) compared with the control diet fed animals. However, the increase in insulin secretion was significantly (P < 0.01) higher in HFD UCP2 (−/−) mice compared with HFD WT mice (Fig. 1C). Similar results were found with fed insulin values (Fig. 1D).

Control diet fasting plasma FFAs were not different between UCP2 (−/−) mice and WT mice. HFD WT mice had significantly (P < 0.05) higher fasting FFA concentrations than their control diet counterparts; however, UCP2 (−/−) mice did not (Table 2). Fed plasma FFA concentrations of control diet fed mice were not different; however, HFD-fed plasma FFAs were significantly elevated in both groups. Plasma triglycerides were not significantly different from each other in control diet fed mice. However, HFD plasma triglycerides were only significantly elevated in HFD WT mice compared with control diet mice (Table 2). Immunoreactive plasma glucagon levels in the control diet and HFD mice were not different between the two groups. HFD WT and UCP2 (−/−) mice had significantly increased plasma glucagon concentrations (P < 0.05; Table 2).

Mice were assessed for glucose tolerance after an IPGTT. UCP2 (−/−) mice had significantly (P < 0.05 to P < 0.001) lower blood glucose concentrations at 10, 20, and 30 min after glucose load than WT mice on HFD (Fig. 2A). The AUC for the blood glucose responses of control diet mice was significantly lower for UCP2 (−/−) mice than WT mice (277.8 ± 15.8 vs. 406.1 ± 13.5 mmol · l⁻¹ · 120 min⁻¹, respectively; P < 0.001). Of importance, for HFD mice, all time points from 10 to 120 min after the glucose load were significantly (P < 0.01) lower for UCP2 (−/−) mice compared with WT mice (Fig. 2B). HFD UCP2 (−/−) mice had a 32% lower AUC than WT mice (526.1 ± 53.6 vs. 789.1 ± 55.6 mmol · l⁻¹ · 120 min⁻¹, respectively; P < 0.05).

The plasma insulin response was also assessed during the IPGTT. Insulin concentrations were measured at 0, 10, 30, and 120 min after glucose challenge. Control diet UCP2 (−/−) insulin responses were significantly elevated at 0, 10, and 30 min compared with control diet WT mice (Fig. 2C). Control diet UCP2 (−/−) mice had significantly higher AUC for 0–30 min than WT mice (24.3 ± 1.6 vs. 14.5 ± 1.2 ng insulin/30 min, respectively; P < 0.05). A similar trend was found in the HFD animals, but the peak insulin response was further exaggerated in the UCP2 (−/−) mice (Fig. 2D). For insulin, UCP2 (−/−) mice had a twofold higher AUC for 0–30 min than WT mice (42.2 ± 6.0 vs. 21.4 ± 2.3 insulin/30 min, respectively; P < 0.05).

Given that UCP2 (−/−) mice had markedly elevated plasma insulin levels, one might expect that they could develop insulin resistance, particularly when fed an HFD. To test this possibility and whether changes in the glucose response to the IPGTT were due to changes in insulin sensitivity, we performed an ITT. Control diet mice showed no difference in their ITT responses (Fig. 2E). However, HFD WT mice were significantly more insulin-resistant than UCP2 (−/−) mice (Fig. 2F).

For assessing β-cell glucose responsiveness in control diet and HFD animals, glucose was presented to the isolated pancreas as a gradient (1.4–17.4 mmol/l; Fig. 3). The control diet UCP2 (−/−) mice showed a significant increase in their glucose sensitivity when compared with control mice (Fig. 3A). The HFD-fed WT mice had a rightward shift in their glucose sensitivity as compared with control diet WT mice (Fig. 3A and B); however, this did not occur in the UCP2 (−/−) mice.

On control diet, pancreatic insulin content was not different between WT and UCP2 (−/−) mice (Table 2). On the HFD, UCP2 (−/−) mice had significantly higher pancreatic insulin content (P < 0.001; Table 2) and islet insulin content corrected for cell number (P < 0.01) than WT mice (Fig. 4A). The glucagon content of the pancreas was not different for control diet or HFD mice (Table 2). To examine the possible mechanism of the increased insulin content, we performed β-cell apoptosis, replication, and islet morphometric analyses. The relative β-cell area of the pancreas was significantly higher in both control diet and HFD-fed UCP2 (−/−) mice (Figs. 4B and 5A and B). HFD UCP2 (−/−) mice had a 2.3-fold higher relative β-cell area compared with HFD control diet mice. There was also an increase in islet number (Fig. 4C) as well as a significant increase in average size of the islets in HFD UCP2 (−/−) mice (Fig. 4D). Apoptosis was two times higher in control diet UCP2 (−/−) mice (P < 0.05; Figs. 4E and 5C and D). After the HFD, apoptosis significantly increased in both WT and UCP2 (−/−) mice (P < 0.001 and P < 0.05, respectively). The percentage of BrdU-positive β-cells was significantly elevated in control diet UCP2 (−/−) mice (Figs. 4F and 5E and F). HFD significantly increased the percentage of BrdU-positive cells in UCP2 (−/−) mice as compared with control diet UCP2 (−/−) mice as well as HFD WT mice.

To examine further the effects of the HFD in mice, we implemented a 48-h in vitro 0.5 mmol/l palmitic acid incubation of islets from either WT or UCP2 (−/−) mice. Before and after FFA treatment, there was no significant difference in islet insulin content among the groups. The WT mice showed elevated basal but attenuated GSIS, whereas UCP2 (−/−) mouse islets showed elevated basal and enhanced GSIS (Fig. 6A and B following palmitate treatment). Islet ATP levels and mitochondrial membrane potential were significantly (P < 0.05) reduced in palmitate-treated WT mouse islets as compared with control islets, whereas UCP2 (−/−) mouse islets showed no significant reduction in ATP levels (Fig. 6C) or mitochondrial membrane potential (Fig. 6D).

To assess the effect of HFD on UCP2 and CPT-1 gene expression in islet RNA, we performed real-time PCR. No detectable UCP2 expression was found in UCP2 (−/−) mice. HFD WT mice, however, had a 6.8-fold increase in UCP2 mRNA compared with their control diet WT counterparts (Table 3). CPT-1 is a key enzyme involved in transporting fatty acids into the mitochondria for β-oxidation and can act, in some situations, to remove lipid accumulation in cells. CPT-1 gene expression was also significantly increased in both WT (P < 0.05) and UCP2 (−/−) mice (P < 0.01) after HFD, but the level of mRNA expression was higher in the UCP2 (−/−) mice (P < 0.05; Table 3).

The detrimental effects of a long-term HFD on β-cell function and insulin sensitivity in mice have been clearly demonstrated and can lead to type 2 diabetes (22,23). The present study focused on comparing the effects of a 4.5-month HFD on glucose homeostasis and pancreatic islet function in WT and UCP2 (−/−) mice. Previous studies have shown that the loss of expression of UCP2 improved glucose tolerance in ob/ob mice; therefore, we hypothesized that UCP2 (−/−) mice would likewise tolerate HFD better than WT mice. We have shown that giving mice an HFD led to elevated plasma triglyceride and fasting FFA concentrations in WT mice, as has been shown in other models of rodent fed diets with elevated fat content (24–26).

As expected, administration of HFD to WT mice was associated with a number of metabolic perturbations, notably insulin resistance and glucose intolerance (Fig. 2). HFD WT mice were more insulin-resistant than HFD UCP2 (−/−) mice. The lack of insulin resistance seen in HFD UCP2 (−/−) mice could be secondary to improved control of diabetes. Although plasma insulin concentrations of HFD WT mice increased in both fasting and postglucose states, the additional secretion was insufficient to maintain a normal glucose profile after an IPGTT. Despite this, HFD UCP2 (−/−) mice maintained lower blood glucose levels than HFD WT mice in both fasting and fed states. It seems that UCP2 (−/−) mice achieved better glucose homeostasis at least partly by a compensatory increase in insulin secretion (Fig. 2D). Therefore, the present data are consistent with a protective effect of low UCP2 on insulin secretion capacity, even when stressed by an HFD. The enhancement seems to be due, in part, to an effect on first-phase insulin secretion in UCP2 (−/−) mice (Fig. 2D). This is further supported in the perfused pancreas model, which shows the maintenance or a small enhancement of glucose sensitivity (Fig. 3).

One mechanism by which pancreatic islet function was preserved in the UCP2 (−/−) mice on HFD was an increase in their insulin secretory capacity. Islet insulin content was significantly higher in HFD UCP2 (−/−) mice than in HFD WT mice. Thus, despite overall higher secretion in vivo, the UCP2 (−/−) mice are able to maintain a sufficient insulin-secretory response when subjected to a long-term HFD. Saturated FFA (the major FFA in lard) inhibits insulin biosynthesis in vitro (27–31).

The total β-cell mass of HFD UCP2 (−/−) mice was also increased; this likely allowed HFD UCP2 (−/−) mice to maintain glucose responsiveness, whereas this did not occur in WT mice. Glucose has been proposed to be one stimulus for increasing pancreatic β-cell mass. Short-term hyperglycemia can stimulate β-cell proliferation, decrease apoptosis, and promote β-cell neogenesis (32–35). The number of functionally intact β-cells in the islet is important for the development, course, and outcome of diabetes. The total β-cell mass reflects the balance between the renewal and loss of these cells (36). Several studies in vitro have so far confirmed the mitogenic action of a high extracellular glucose concentration (32–35,37). However, there is no explanation of how glucose stimulates these changes or how β-cells lose this ability during chronic hyperglycemia. The current data suggest that hyperglycemia per se is not essential for stimulating β-cell replication or proliferation. Instead, enhanced glucose metabolism permitted by higher levels of coupled oxidative phosphorylation may be the stimulus for β-cell proliferation and neogenesis. Thus, the expression of UCP2 may limit the ability of the β-cell to compensate for increasing insulin resistance by increasing β-cell mass. The mechanism by which UCP2 may perform this task is to limit the production of a metabolic stimulus, such as ATP, for increasing β-cell mass. ATP, for example, could be working through the mammalian target of rapamycin to stimulate β-cell proliferation (38). Work is currently under way in our laboratory to explore further this potential relationship.

CPT-1 activity is critical to promoting FFA acid oxidation in cells; moreover, its expression is induced in β-cells by FFA exposure (39). Preventing transport of FFAs to the mitochondrial matrix by use of CPT-1 inhibitors prevents the inhibition in GSIS observed after exposure of islets to long-chain FFAs (28). Although gene expression levels do not necessarily equate to functional activity, it is of interest to point out that UCP2 (−/−) mice fed control diet had elevated CPT-1 mRNA expression compared with WT islets. After HFD, the difference increased. The implication of this finding is that islets from UCP2 (−/−) mice may be able to oxidize greater amounts of FFA compared with WT mice. This would prevent accumulation of islet triglyceride, which is a key marker of lipotoxicity (40). CPT-1 induction could be the dominant mechanism by which islets from UCP2 (−/−) mice maintain their high level of functionality under stress. Alternatively, it could be a complementary mechanism to our primary hypothesis: that maintenance of higher levels of ATP (whether via glucose or fat metabolism) is key to enhanced β-cell glucose responsiveness and proliferative capacity. Additional studies, such as the measurement of the ATP/ADP ratio and a comprehensive assessment of glucose and FFA metabolism, are required to quantify the relative contributions to islet health from these two possible outcomes in UCP2 (−/−) mice. It is also important to note that UCP2 and CPT-1 are coordinately upregulated by FFA in WT mice, yet the upregulation of CPT-1 alone does not prevent the loss of β-cell function. Moreover, UCP2 is downregulated in β-cells exposed to high glucose, whereas CPT-1 expression remains unchanged (41); therefore, simply examining the ratio between the two entities does not predict the outcome on β-cell metabolism.

The concept of lipotoxicity in β-cells is attractive as an explanation for the loss of GSIS in obesity-induced diabetes. However, evidence is accumulating that fats exert most of their deleterious effects on islets only when glucose is also elevated; this concept is termed “glucolipotoxicity” (42). Thus, only in hyperglycemic rats did HFD cause a loss of GSIS concomitant with induction of UCP2 (43). Our current results are similar because UCP2 (−/−) mice that maintained fasting euglycemia were also resistant to HFD. It is interesting that in the Briaud et al. study (43), HFD did not increase islet triglyceride, suggesting that fat accumulation is not a necessary consequence of exposure of islets to high fat.

As a consequence of its ability to dissipate the proton-motive force, UCP2 is predicted to reduce the generation of ROS (44). This hypothesis is supported by data showing that macrophages from UCP2 (−/−) mice produce more ROS when challenged with infection (12) and that UCP2 is activated by ROS in mitochondria from MIN6 cells (7). An increase in ROS can lead to β-cell apoptosis (45); therefore, the presence of elevated UCP2 may reduce ROS production by dissipating the excess energy. This excessive energy metabolism, as seen with lipid exposure, may induce multiple mechanisms that lead to cell damage, such as ceramide formation (46) and apoptosis (47) in islets. At the outset, it was unclear whether the positive effects of the lack of UCP2 on insulin secretion would be negated by the putative greater increase in apoptosis in HFD UCP2 (−/−) mice. However, this was not the case, at least in the time frame studied here. These results also suggest that inhibition and not activation of UCP2 could be an effective means of improving β-cell function in individuals with type 2 diabetes. Physiologically, an advantage of increasing the expression of UCP2 may be to prevent long-term effects of excessive fat expression often associated with increasing ROS production; however, as we have shown, UCP2 negatively impacts insulin secretion and the compensatory increase in pancreatic β-cell mass.

In conclusion, UCP2 (−/−) mice have enhanced β-cell responsiveness to a glucose challenge in vivo after an HFD. These mice also have preserved islet sensitivity to glucose, enhanced insulin content, and increased β-cell mass after an HFD. These factors seem to contribute to an increased insulin response to glucose in vivo that produces improvements in glucose tolerance in the face of insulin resistance.

This work was supported by operating grants from the Canadian Institutes of Health Research (CIHR) to M.B.W. and C.B.C. (MOP 12898 and MOP 85464). J.W.J. was supported by the Banting Best Diabetes Center studentship award and by a CIHR doctoral award. M.B.W. is supported by a CIHR investigator award. C.-Y.Z. and B.B.L. are supported by research grants from the National Institutes of Health and the American Diabetes Association.

We thank Dr. Sandy Der for assistance with real-time PCR studies.