Differential Effects of Hyperlipidemia on Insulin Secretion in Islets of Langerhans From Hyperglycemic Versus Normoglycemic Rats

The spontaneously diabetic GK rat is a lean model of type 2 diabetes characterized by deficient insulin response to glucose in vivo and in vitro (15). Our colony originates from that of Dr. R. Farese (Tampa, FL) and is maintained at the Pacific Northwest Research Institute (PNRI). Age-matched male Wistar rats were purchased from Harlan Teklad. GK and Wistar rats 6 weeks of age were caged with free access to water and fed either a high-fat diet (Harlan Teklad, Madison, WI) or a regular diet for 6 or 10 weeks ad libitum. The composition of both diets is given in Table 1. Both were adequately supplemented with vitamins and minerals. In the high-fat diet, the relative proportions of FAs were 39.1% saturated, 48.9% monounsaturated, and 12.1% polyunsaturated. Animals were housed on a 12-h light/dark cycle (9:00 p.m. to 9:00 a.m. dark cycle). All procedures using animals were approved by the PNRI Institutional Animal Care and Use Committee.

At the end of the diet, overnight-fasted GK and Wistar rats were given 2 g/kg glucose orally by gavage using a 60% glucose solution. Blood was collected from the tail vein at 0, 30, 60, 90, and 120 min after dosing. Plasma glucose levels were determined using a Glucose Analyzer 2 (Beckman Instruments, Fullerton, CA), and plasma C-peptide levels were measured by a C-peptide radioimmunoassay (RIA) kit (Linco Research, St. Louis, MO).

Epididymal fat was excised and weighed. Intracardiac blood samples were collected into EDTA-containing vials. Plasma was immediately separated by 3,000-rpm centrifugation at 4°C for 10 min. Plasma TG and FA levels were measured immediately using GPO-Trinder reagent (Sigma Diagnostics, St Louis, MO) and the Half-Micro test (Boehringer-Mannheim, Roche Diagnostics, Mannheim, Germany), respectively. Plasma insulin and leptin levels were measured using the Sensitive Rat Insulin RIA kit and the Rat Leptin RIA kit, respectively (Linco Research).

Rat islets were isolated from nonfasted animals by collagenase digestion as described (16). After isolation, islets were cultured for 1 h in RPMI 1640 containing 10% fetal bovine serum and 11.1 mmol/l glucose. Batches of 10 islets each were washed twice in Krebs-Ringer buffer (KRB) containing 0.1% BSA and 2.8 mmol/l glucose for 20 min at 37°C, and then they were incubated for 60 min in the presence of 2.8, 5.6, 8.3, 11.1, or 16.7 mmol/l glucose; 40 mmol/l KCl; 10 mmol/l dl-glyceraldehyde; or 20 mmol/l succinic acid monomethyl ester, as indicated in the figure legends. Each condition was run in duplicate. Insulin levels in samples collected from the static incubations were measured using the Sensitive Rat Insulin RIA kit (Linco Research).

Comparative analysis of insulin mRNA levels was performed using the Gold RT-PCR kit from Perkin-Elmer and an ABI Prism 7700 sequence detector equipped with a thermocycler (Taqman technology), as previously described (12). β-Actin was used as an internal control. Primer and probe sequences for insulin and β-actin have been described elsewhere (12). Insulin content was measured in duplicate batches of 20 islets after acid-alcohol extraction. TG content was measured in batches of 100 islets after chloroform:methanol extraction as previously described (12).

Batches of 100 isolated islets each were preincubated twice for 20 min at 37°C in KRB containing 0.1% BSA, and then they were incubated in KRB containing 0.1% BSA, either 2.8 or 16.7 mmol/l unlabeled glucose, and 0.1 μCi/μmol [U-¹⁴C]glucose (NEN, Boston, MA) for 1 h at 37°C. A volume of 1 ml incubation buffer was then transferred to an Erlenmeyer flask with a rubber cap and acidified with 100 μl of 7% perchloric acid. Benzethonium hydroxide (400 μl) was then injected into small wells suspended under the rubber caps, and after 16 h at room temperature, the trapped ¹⁴CO₂ was measured by liquid scintillation counting. Background counts obtained from a control batch treated side by side with the samples but without islets was subtracted from the counts.

Batches of 100 islets each were harvested in 75 μl lysis buffer (140 mmol/l NaCl, 10 mmol/l Tris-HCl, pH 7.4, 1 mmol/l CaCl₂, 1 mmol/l MgCl₂, 10% glycerol, 1% Nonidet P-40, 1 mmol/l Na₃VO₄, 0.5 mmol/l phenylmethysulfonyl fluoride, 0.5 mmol/l dithiothreitol, and 1 μg/ml each of leupeptin, aprotinin, antipain, and pepstatin A), frozen and thawed three times, and cell debris was removed by centrifugation. A volume of 30 μg protein were separated by SDS-PAGE with 12% acrylamide gels. Resolved proteins were transferred to nitrocellulose membranes (BioRad, Hercules, CA) by electroblotting. For immunoblotting, the membranes were blocked with 1% nonfat dry milk in PBS with 0.05% Tween (PBS-T) for 1 h and then incubated for 1 h at room temperature with a rabbit polyclonal antibody raised against the COOH terminus of mouse UCP-2 (Alpha Diagnostic International, San Antonio, TX) as previously described (13). Membranes were then washed in PBS-T and incubated with horseradish peroxidase-coupled rabbit IgG antibody (Santa Cruz Biotechnologies, Santa Cruz, CA) for 1 h at room temperature. The bands were visualized by enhanced chemiluminescence detection reagents (Amersham, U.K.) and exposed to X-Omat Blue Kodak films for 1–5 min. Membranes were stripped and reprobed with an anti-bovine cytochrome oxidase subunit IV mouse monoclonal antibody (Molecular Probes, Eugene, OR).

A group of GK rats was treated for 6 weeks, during the diet period, with either subcutaneous insulin implants releasing 2 units/24 h of insulin (Linplant, Linshin Canada, ON, Canada) or blank implants. Additional insulin pellets were implanted as needed over the course of the study to maintain normal plasma glucose levels. Fed glucose levels from plasma samples collected from the tail vein were determined once a week using an Accu-Chek device (Roche).

Data are expressed as the means ± SE. Intergroup comparisons were performed using Student’s unpaired t test or ANOVA with post hoc Bonferonni adjustment, where appropriate. P < 0.05 was considered significant.

Metabolic parameters measured after 6 weeks of diet are shown in Table 2. GK rats fed a regular diet had higher plasma glucose (P < 0.05, n = 12), lower body weight (P < 0.05, n = 12), lower epididymal fat weight (P < 0.05, n = 8), and lower plasma TG levels (P < 0.05, n = 8) than Wistar rats. Plasma FA levels were similar in GK and Wistar rats (NS, n = 8). As expected, epididymal fat weight, plasma TG, and FA levels were increased after 6 weeks of high-fat diet in both GK and Wistar rats. Fasting plasma glucose levels were slightly increased by the high-fat diet in both groups, but the difference was not statistically significant. Fed insulin levels were significantly increased by high-fat diet in Wistar (P < 0.05, n = 19–22) but not in GK rats.

To further explore the effects of a high-fat diet on glucose tolerance and insulin secretion, oral glucose tolerance tests were performed after 6 weeks of diet (Fig. 1). As previously reported (17), glucose tolerance was decreased in GK rats compared with Wistar rats (Fig. 1A) and was accompanied by markedly decreased circulating C-peptide levels (area under the curve [AUC] 16,572 ± 7,025 vs. 91,472 ± 13,519 pmol · l⁻¹ · 120 min⁻¹, n = 6, P < 0.001) (Fig. 1B). Glucose tolerance was slightly but significantly decreased by high-fat diet in both Wistar (n = 10, P < 0.05, ANOVA) and GK (n = 12, P < 0.05, ANOVA) rats (Fig. 1A). The C-peptide response was not significantly increased by high-fat diet in either Wistar (AUC 122,762 ± 16,723 vs. 91,472 ± 13,518 pmol · l⁻¹ · 120 min⁻¹, n = 6, NS) or GK (27,872 ± 5,151 vs. 16,572 ± 7,025 pmol · l⁻¹ · 120 min⁻¹, NS, n = 6) rats (Fig. 1B). Similar effects of the diet on glucose tolerance were observed after 10 weeks (data not shown).

To determine the effects of high-fat diet on insulin secretion in vitro, glucose-induced insulin release was measured in static incubation in islets isolated after 6 weeks of diet (Fig. 2A). As previously reported (15), glucose-induced insulin secretion was impaired in GK rat islets as compared with Wistar rat islets. Insulin secretion was essentially unchanged in islets from Wistar rats fed a high-fat diet compared with those fed a regular diet. In contrast, insulin secretion induced by 16.7 mmol/l glucose was decreased by ∼50% in islets from high fat-fed GK rats compared with regular diet-fed GK rats (36 ± 7 vs. 69 ± 7 μU/islet, n = 6, P < 0.05). Similar effects were observed after 10 weeks of diet (data not shown).

To further characterize the insulin secretory defect observed in islets from high fat-fed GK rats, we measured insulin secretion induced by secretagogues acting at different steps of the stimulus/secretion-coupling pathway. The response to glucose was decreased after 6 weeks of high-fat diet in islets from GK rats (26 ± 3 vs. 52 ± 12 μU/islet, n = 7, P < 0.05) (Fig. 2B), confirming the results shown in Fig. 2A. Insulin secretion induced by either glyceraldehyde (a glycolytic intermediate) or succinic acid monomethyl ester (an intermediate of the Krebs cycle) was decreased in islets from GK rats compared with Wistar rats (glyceraldehyde: 20 ± 5 vs. 65 ± 11 μU/islet; n = 6, P < 0.05; succinic acid monomethyl ester: 17 ± 5 vs. 52 ± 12 μU/islet, n = 6, P < 0.05) (Fig. 2B). Insulin secretion induced by KCl was not statistically decreased in GK rat islets compared with Wistar rat islets (23 ± 5 vs. 36 ± 7 μU/islet, n = 6, NS) (Fig. 2B). None of the responses to nonglucose secretagogues was affected by high-fat feeding (Fig. 2B).

To determine whether the impairment of glucose-induced insulin secretion was caused by a decrease in intracellular insulin stores, we measured insulin content and insulin mRNA levels in islets isolated after 6 weeks of high-fat diet. As previously reported in the same colony of GK rats (18), insulin content in islets from GK rats was similar to that of control rats (1,676 ± 124 vs. 1,831 ± 158 μU/islet, n = 6, NS). Insulin content was not significantly modified by the high-fat diet in islets from GK rats (1,386 ± 149 vs. 1,676 ± 124 μU/islet, n = 6, NS) or Wistar rats (1,884 ± 238 vs. 1,831 ± 158 μU/islet, n = 6, NS). Insulin mRNA levels were markedly decreased in islets from GK rats compared with Wistar rats (1.7 ± 0.5 vs. 7.8 ± 2.1, P < 0.05, n = 6–8) but were not affected by high-fat diet in either Wistar (7.8 ± 2.2 vs. 7.8 ± 2.1, n = 8, NS) or GK (2.2 ± 0.7 vs. 1.7 ± 0.5, n = 6, NS) rat islets. Similar results were observed after 10 weeks of diet (data not shown).

Insulin was administered in the form of subcutaneously implanted pellets to high fat-fed and regular diet-fed GK rats for 6 weeks. Metabolic parameters measured after diet and insulin treatment are shown in Table 3. In contrast to observations made with the first group of rats (Table 2), high-fat feeding significantly increased fasting plasma glucose levels in GK rats (n = 6, P < 0.05) (Table 3). As expected, insulin treatment lowered plasma glucose levels in GK rats to levels comparable to those of Wistar rats. In insulin-treated GK rats, high-fat feeding did not increase plasma FA levels, whereas plasma TG levels were significantly increased. Plasma leptin levels were significantly increased by high-fat feeding (n = 6, P < 0.0001, ANOVA) (Table 3) but were not affected by insulin treatment. Glucose tolerance was increased by insulin treatment, but the effect was significant only in high fat-fed rats (AUC 12,412 ± 2,049 vs. 25,710 ± 3,095 mg/ · dl⁻¹ · 120 min⁻¹, n = 6, P < 0.01). In isolated islets, insulin secretion induced by 16.7 mmol/l glucose was markedly decreased in the high fat-fed group (23 ± 3 vs. 65 ± 7 μU/islet, n = 9, P < 0.05) (Fig. 3), confirming our results shown in (Fig. 2. Additionally, potentiation of glucose-induced insulin secretion by palmitate was decreased in islets from high fat-fed GK rats Fig. 3). Importantly, glucose and palmitate stimulation of insulin secretion was fully restored by insulin treatment (Fig. 3). Insulin content was not significantly different between the four groups of animals (data not shown).

To investigate the mechanisms of high-fat diet-induced impairment of insulin secretion in GK rat islets, we measured TG content in islets after 6 weeks of high-fat feeding. Islet TG content was similar between GK and Wistar rats and was not affected by the diet in either group (Wistar rats: 128 ± 24 ng/islet in high fat-fed animals vs. 150 ± 43 ng/islets in regular diet-fed animals, n = 6, NS; GK rats: 127 ± 33 ng/islet in high fat-fed animals vs. 194 ± 35 in regular diet-fed animals, n = 6, NS).

Neither high-fat feeding nor insulin treatment modified the rate of glucose oxidation at either 2.8 or 16.7 mmol/l glucose in islets from GK rats (Table 4). We compared UCP-2 protein levels by Western analysis in islets isolated from GK rats after 6 weeks of high-fat feeding with or without concurrent insulin treatment (Fig. 4). UCP-2 protein levels were increased 2.0 ± 0.4-fold (P < 0.05, n = 3) after high-fat feeding in islets from untreated GK rats. In marked contrast, UCP-2 protein levels did not increase in islets from high fat-fed rats treated with insulin (0.4 ± 0.2-fold as compared with islets from insulin-treated, regular diet-fed GK rats, n = 3, NS). These changes were specific for UCP-2 because expression of the mitochondrial protein cytochrome oxidase was not affected by high-fat feeding (1.0 ± 0.1-fold increase, n = 3, NS) or insulin treatment (1.0 ± 0.1-fold increase, n = 3, NS). These results indicate that the deleterious effects of high-fat feeding on insulin secretion in GK rat islets are associated with increased UCP-2 expression.

This study was designed to assess whether hyperlipidemia induced by high-fat feeding differentially affects β-cell function in hyperglycemic versus normoglycemic rats, and to determine whether the mechanisms of these effects involve TG accumulation, changes in glucose oxidation, or increased UCP-2 expression in islets. Our results show that high-fat diet impairs glucose-induced insulin secretion in islets from GK rats but not from Wistar rats and that this is associated with increased UCP-2 expression. Both the decrease in glucose-induced insulin secretion and the increase in UCP-2 protein levels are prevented by insulin treatment.

Variable effects of a high-fat diet on insulin secretion have been reported in the literature (19–22). Ahrèn et al. (19) have shown that glucose-induced insulin secretion in islets from normal rats was decreased after 4 weeks of high-fat diet. This was associated with a decreased insulin response to glucose in vivo. Others (20,21) have observed an increase in plasma insulin levels attributed to insulin resistance in normal rats fed a high-fat diet, whereas in another study, Kahn and Pedersen (22) showed that high-fat diet decreases plasma insulin levels. These discrepancies may be explained by variations in the composition of the diet and in the balance between the effects of lipids on insulin sensitivity and on insulin secretion. In the present study, high-fat feeding for up to 10 weeks increased circulating insulin levels but did not impair glucose-induced insulin secretion in vivo or in vitro in Wistar rats. In contrast, the maximal response to glucose was markedly impaired in islets from high fat-fed GK rats. Although we did not directly measure the effects of high-fat feeding on β-cell mass, the fact that basal insulin release, insulin content, and insulin mRNA levels were not affected by the diet indicates that the decreased insulin response to glucose is a true functional impairment rather than an indirect effect caused by changes in β-cell mass. Insulin secretion in response to glyceraldehyde, succinate methyl ester, and KCl were unchanged after 6 weeks of high-fat diet, suggesting that high-fat diet-induced hyperlipidemia selectively affects the response to glucose, a characteristic also observed in human type 2 diabetes (23). An alternative explanation is that any impairment in the response to glyceraldehyde, succinate methyl ester, and KCl in islets from high fat-fed GK rats may not be detectable, given the already small amplitude of the response observed in islets from normal diet-fed GK rats. Potentiation of glucose response in the presence of the FA palmitate was also decreased by high-fat feeding in GK islets. Insulin treatment did not modify the response to glucose or palmitate in islets from regular diet-fed rats, but it completely prevented the inhibitory effect of high-fat diet on insulin secretion. Because both glucose and FA levels were normalized by insulin treatment, we cannot unequivocally conclude that the restoration of insulin secretion was due to normalization of glycemia. However, it is noteworthy that prevention of the high-fat diet-induced insulin secretory defect in GK rat islets by insulin treatment was not associated with decreased adiposity (as assessed by epididymal fat weight), circulating TG levels, or plasma leptin concentrations, confirming that effects of high-fat feeding are cause by the synergistically deleterious association of hyperglycemia and hyperlipidemia.

In ZDF rats, the marked increase in islet TG content associated with hyperlipidemia has been suggested to play a mechanistic role in β-cell dysfunction and loss (2), mainly by inducing apoptosis, the primary cause of decreased β-cell mass in this model (24). Marked elevation in islet TG content and deposition of fat droplets has also been observed in OLETF rats, a model of hypertriglyceridemia-associated diabetes (25). Here we show that impairment of glucose-induced insulin secretion by high-fat diet in hyperglycemic GK rats is not associated with a detectable increase in islet TG content. Recently, Lee et al. (7) observed that leptin deficiency in ZDF rats results in the ectopic accumulation of TG in nonadipose tissues, and they have suggested that overnutrition-induced hyperleptinemia protects nonadipose tissues from lipid accumulation. This antisteatotic role of hyperleptinemia may account for the fact that in leptin-sensitive GK as well as Wistar rats, high-fat diet did not lead to islet TG accumulation. The present results contrast with our previous observations in isolated rat islets that prolonged exposure to elevated FAs in the presence of high glucose is associated with increased TG content and decreased insulin gene expression (9,12). We speculate that these discrepancies might be caused by differences between in vitro and in vivo levels of FAs, as well as a different sensitivity of isolated islets in vitro and in situ islets in vivo to the effects of FAs. Regardless of the cause for this discrepancy, the present results indicate that the concomitant presence of hyperglycemia and hyperlipidemia can impair glucose-induced insulin secretion without significant changes in insulin gene expression.

Several studies have reported that chronic exposure to elevated levels of FAs induces an increase in FA oxidation and a decrease in glucose oxidation (26–30), suggesting that a glucose-FA cycle, as described in the muscle (31), may be operative in the β-cell. In contrast, other groups have found minimal, if any, changes in glucose oxidation after culture of β-cell lines (32,33) or islets (34) in the presence of FAs. Similarly, in the present study, we did not find any effect of high-fat feeding on glucose oxidation, suggesting that the site of action of FAs lies downstream of mitochondrial oxidation.

UCPs are located in the inner mitochondrial membrane, and they uncouple mitochondrial respiration from oxidative phosphorylation, resulting in decreased ATP production (35). UCP-2 is a ubiquitously expressed UCP family member, and its expression is increased by high-fat feeding or FAs in various tissues (35,36). Adenovirus-mediated overexpression of UCP-2 in islets decreases glucose-stimulated insulin secretion (14,37). Recently, it has been suggested that an increase in UCP-2 expression might mediate deleterious effects of FAs on β-cell function in vitro (13) and in vivo (38). We found increased UCP-2 protein levels in islets from GK rats fed a high-fat diet, and this increase was prevented by insulin treatment. Whether these changes are due to translational or transcriptional regulation of UCP-2 remains to be determined. Although these results do not prove that the impairment in insulin secretion was caused by changes in UCP-2 expression, they are consistent with a model in which elevated plasma levels of FAs increase UCP-2 levels and mitochondrial uncoupling, thereby decreasing ATP generation and thus insulin secretion, without changes in glucose oxidation. This model does not exclude the possibility that the expression of other metabolic genes may be modulated by the association of hyperglycemia and hyperlipidemia. Indeed, both of these abnormalities are known to modify the expression of a number of genes and metabolic pathways implicated in glucose-stimulated insulin secretion (rev. in 8). Understanding the relative contribution of changes in UCP-2 expression and other potential targets requires further studies.

In conclusion, our results demonstrate that high-fat feeding impairs glucose-induced insulin secretion only in the context of chronic hyperglycemia in GK rat islets. These effects are not mediated by TG accumulation in islets but are associated with changes in UCP-2 expression. Whether similar mechanisms prevail in obesity-associated diabetes in humans remains to be established.

This work was supported in part by a research grant from the American Diabetes Association (to V.P.) and a fellowship from l’Aide aux Jeunes Diabétiques (to I.B.).

We are grateful to Drs. R. Paul Robertson and Christopher J. Rhodes for careful reading of the manuscript and to Madeline Johnson for excellent secretarial assistance.