Impaired Glucose-Stimulated Insulin Secretion Is Coupled With Exocrine Pancreatic Lesions in the Cohen Diabetic Rat

Cohen diabetic rats were bred and maintained in the animal facility at the Hebrew University School of Medicine, Jerusalem. Rats were fed regular diet ad libitum (Koffolk, Petach-Tikva, Israel), composed of 54% carbohydrate (ground whole wheat, ground alfalfa, and bran); 21% protein (skimmed milk powder); 6% fat; 5% salts, vitamins, and trace elements including an adequate copper content (15 ppm); 7% humidity; and 7% ash. Custom-prepared high-sucrose, copper-poor diet contains 72% sucrose; 18% vitamin-free casein; 5% salt mixture no. II USP (MP Biomedicals, Solon, OH); 4.5% butter; and 0.5% corn oil, vitamins and low copper (0.9 ppm) (6,7). Animal studies were approved by the institutional committee for animal use and care.

Seven-week-old male rats on regular diet were switched to high-sucrose, copper-poor diet for an additional period of 1 month. Two groups were studied: Cohen diabetes–sensitive and Cohen diabetes–resistant rats fed high-sucrose, copper-poor diet (CDs-HSD and CDr-HSD, respectively). Care was taken to include in each study Cohen diabetes–sensitive and Cohen diabetes–resistant rats with similar weights to preclude possible influence of the initial weight on the phenotype. In vivo studies included postprandial determination of glucose and insulin, oral or intravenous glucose tolerance test (OGTT or IVGTT, respectively), insulin tolerance test (ITT), and insulin secretion in response to non-nutrient secretagogues. Glucose concentration was measured in tail blood using a standard glucometer (Elite; Bayer, Leverkusen, Germany). Serum insulin was determined using an ultrasensitive rat insulin Elisa assay (Mercodia, Uppsala, Sweden). In vitro studies comprised insulin secretion in perfused pancreas and proinsulin biosynthesis and insulin secretion in isolated islets.

Rats were fasted overnight. Blood glucose and serum insulin concentrations were measured after fasting and after 60 min of free access to high-sucrose, copper-poor diet. Additional blood samples were taken at 60 and 120 min after high-sucrose, copper-poor–diet removal.

Blood glucose and serum insulin concentrations were measured after overnight fast and after the oral administration of 3.5 g/kg glucose, as described previously (6).

Heparinized catheters were inserted into the right jugular vein and left carotid artery of anesthetized rats (85 mg/kg Ketalar [Parke-Davis, Gwent, U.K.] and 3 mg/kg xylazine [XYL-M2 Veterinary; VMD, Arendonk Belgium]) 7 days before tests.

Blood glucose and serum insulin concentrations were measured after overnight fast and during 120 min after a bolus administration of 0.75 g/kg glucose via the jugular vein.

Bolus arginine (350 mg/kg) or saline followed by 45 min of 6 mg/min constant infusion was administered intravenously to CDs-HSD rats after an overnight fast or in the postprandial state. Tolbutamide (bolus of 100 mg/kg i.v.) was administered after an overnight fast. Carotid artery blood glucose and serum insulin concentrations were measured before and after arginine, tolbutamide, or saline administration (control). Control and tests were performed on the same rats after 4–5 days of recovery.

Blood glucose concentrations were measured in CDs-HSD and CDr-HSD rats after an overnight fast and after intraperitoneal administration of insulin in three doses: 0.25, 0.5, and 1 unit/kg (Actrapid HM; Novo Nordisk, Bagsvaerd, Denmark).

Rats were anesthetized. Liver, spleen, kidney, heart, and pancreas were removed, dissected from external fat, and weighed. The entire pancreas was flushed-frozen in liquid nitrogen and kept at −80°C for insulin or triglyceride (TG) determination. Alternatively, the pancreas was dissected and immersed in electron microscopy fixative solution or fixed in formalin for further histological analyses.

Insulin was extracted from the entire pancreas with acid-ethanol solution (21) and assayed by radioimmunoassay (RIA) using a commercial kit (Linco Research, St. Charles, MO).

TGs were extracted from the entire pancreas using the Folch method (22) and determined using the GPO-Trinder kit (Sigma, St. Louis, MO).

Serum free fatty acids (FFAs) were measured in sera of overnight fasted CDs-HSD and CDr-HSD rats using the NEFA kit (Randox Laboratories, Crumlin, County Antrim, U.K.).

Paraffin-embedded pancreatic sections (5 μm) stained with hematoxylin-eosin were evaluated for islet morphology, macrophage infiltration, and acinar degeneration (at Hadassah Hospital, Jerusalem, and Hannover Medical School). For immunohistochemical analysis, sections were incubated overnight with guinea pig anti-insulin antibody (1:500) (polyclonal A565; DAKO, Hamburg, Germany), mouse anti-glucagon antibody (1:300) (monoclonal; Sigma), mouse anti-ED1 (1:500) (monoclonal MCA 341), or mouse anti-IL-1β (1:200) (monoclonal AAR15G) (Serotec, Oxford, U.K.) antibodies followed by a 30-min incubation with an appropriate biotinylated second antibody, as described previously (23) and a 30-min incubation with a mixture of streptavidin (1:100) and biotin-peroxidase (1:1,000) (Jackson ImmunoResearch Laboratories, West Grove, IL) (24). The peroxidase reaction was visualized using 0.7 mmol/l diaminobenzidine and 0.002% hydrogen peroxide in PBS, pH 7.3. Sections were examined by bright-field illumination or under phase contrast using a Zeiss Photomicroscope II (Zeiss, Oberkochen, Germany).

Thin pancreas sections fixed overnight in cacodylate buffer (pH 7.3), containing 2% glutaraldehyde and para-formaldehyde were stained with 1% osmium tetraoxide (lipids stained black), dehydrated, and embedded in Epon. Sections were contrast-stained with saturated solutions of lead citrate and uranyl acetate and evaluated by electron microscopy (EM 9 S2; Zeiss) (23).

Sections (400 μm) of formalin-fixed pancreas and gastrocnemius muscle were stained with 0.1% osmium tetraoxide followed by mild staining (0.05%) with uranyl acetate (lipids appear bright white). Stained samples were kept at 4°C. Scanning electron microscopy (SEM) imaging was performed on samples placed in a capsule (QuantomiX, Rehovot, Israel) using SEM (FEI XL-30; FEI, Eindhoven, The Netherlands) (25).

Pancreases were removed 4 h after 100 mg/kg 5-bromo-2′-deoxyuridine (BrdU; Sigma) intraperitoneal injection. Cell replication was assessed by scoring BrdU staining (incorporation) of proliferating cell nuclei (Biosciences Pharmingen, Brussels, Belgium). Apoptotic cells were identified by terminal dUTP 3′ nick end-labeling (TUNEL) procedure using cell death detection kit (Roche, Mannheim, Germany) (26). Stained nuclei were counted and expressed as percentage of the total number of nuclei in the same field (23,27).

Rats were anesthetized (100 mg/kg pentobarbital). The pancreas was dissected and perfused (3 ml/min) with Krebs-Ringer bicarbonate buffer (KRBB) in a 37°C chamber via the abdominal aorta. Subsequently, pancreases were perfused with KRBB containing 3.3 mmol/l glucose for 20 min; 16.7 mmol/l glucose for 60 min, and 3.3 mmol/l glucose for additional 20 min. Samples were collected from the pancreatic vein and stored at −20°C for insulin RIA (Linco) (28).

Batches (200–300) of similar size islets, washed repeatedly with Hank's balanced salt solution, were incubated at 37°C for 90 min under 5% CO₂ atmosphere in modified KRBB containing 20 mmol/l HEPES and 0.25% BSA (KRBH-BSA) and 3.3 mmol/l glucose to allow recovery from the isolation procedure. Groups of 25 islets were washed in KRBH-BSA buffer containing 1.7 mmol/l glucose and incubated for 1 h in the same buffer containing increasing concentrations of glucose (1.7–16.7 mmol/l). Insulin release was determined in supernatants of the incubation buffer. Islets were then suspended in fresh KRBH-BSA buffer containing 25 μCi l-[³H]leucine (150 Ci · mmol⁻¹ · l⁻¹; Amersham, Aylesbury, U.K.) (29) and the same glucose concentration and incubated for additional 15 min at 37°C. Leucine incorporation was terminated by addition of 1 ml ice-cold glucose-free KRBH-BSA buffer and rapid centrifugation. Islet insulin content and proinsulin biosynthesis was determined in pellet extracts suspended in glycine buffer and subjected to four freeze-thaw cycles in liquid nitrogen, as described previously (30). Islet insulin secretion and content were determined using rat insulin RIA kit (Linco).

Data shown are means ± SE. Statistical significance of differences between groups was determined by one-way ANOVA followed by Tukey's test using the Sigmastat program (Jandel, San Rafael, CA). A two-tail paired t test was used to compare data from tests performed on the same animal or in isolated islets. A P value of <0.05 was considered significant.

In CDs-HSD rats, the markedly elevated blood glucose concentrations remained high 120 min after high-sucrose, copper-poor–diet removal (Fig. 1A). The high blood glucose concentrations were coupled with a low-flat insulin output (Fig. 1B). The CDr-HSD rats maintained normal blood glucose levels and exhibited a significant postprandial increase in insulin secretion (Fig. 1A and B).

Glucose-stimulated insulin secretion (GSIS) was assessed in CDs-HSD and CDr-HSD rats in response to oral (Fig. 1C and D) and intravenous (Fig. 1E and F) glucose administration. CDs-HSD rats exhibited an abnormal glucose-tolerance curve, characterized by elevated glucose levels and a low insulin output, whereas CDr-HSD rats exhibited a normal response (Fig. 1C–F). The calculated glucose area under the curve (AUC) of CDs-HSD rats (Fig. 1G) was higher in both OGTT and IVGTT compared with CDr-HSD rats (Fig. 1H), whereas the insulin AUC was lower. The insulin response was higher in OGTT compared with IVGTT in both CDs-HSD (Fig. 1G) and CDr-HSD rats (Fig. 1H). A comparable normal IVGTT was demonstrated for both Cohen diabetes–sensitive and Cohen diabetes–resistant rats maintained for 30 days on regular diet, as shown in the online appendix (Supplemental Fig. 1Sa–b, available at http://dx.doi.org/10.2337/db07-0520).

The mechanism underlying the reduced GSIS of the CDs-HSD rats was examined in response to arginine and tolbutamide. The abnormal postprandial (Fig. 2A and B) or fasting (Fig. 2C and D) glucose and insulin response curves observed in CDs-HSD rats were fully normalized by arginine or tolbutamide administration (Fig. 2A–D). The glucose AUC was significantly decreased whereas the insulin AUC was increased (Fig. 2E and F) compared with control.

Insulin produced a similar decrease in blood glucose levels in both CDs-HSD and CDr-HSD rats (Fig. 3A). The CDs-HSD rats exhibited a dose-dependent reduction in blood glucose levels (Fig. 3B), indicating a high sensitivity to insulin, which may explain the similar basal glucose levels of CDs-HSD and CDr-HSD rats.

The hyperglycemic CDs-HSD rats were nearly 20% smaller than the CDr-HSD rats (230 ± 7 vs. 290 ± 6 g) after 5 weeks on high-sucrose, copper-poor diet but did not show significant body weight difference after 1–3 weeks of high-sucrose, copper-poor diet (data not shown). The pancreas weight of the CDs-HSD rats was 40% less than CDr-HSD pancreas. No such variance was detected in the weight of the liver, spleen, heart, or kidney (Table 1).

Considerable apoptotic activity (2.48 ± 0.27%) was detected in the acinar cells of the CDs-HSD rats using the TUNEL assay, whereas only a few (<0.01%) apoptotic cells were detected in CDr-HSD rats. The acinar cells of the CDs-HSD rats exhibited increased proliferative activity compared with CDr-HSD rats (1.01 ± 0.02 vs. 0.08 ± 0.02% [BrdU labeling index], respectively, P < 0.01). The very few (<0.01%) TUNEL- and BrdU-positive cells detected in the islets of the same animals precluded quantitative analysis. Thus, acinar cell apoptosis is likely to be the cause of in the reduced pancreatic weight of the CDs-HSD rats (Table 1).

Pancreatic insulin content of the CDs-HSD rat was lower but not significantly different compared with CDr-HSD rats (16.8 ± 3.4 vs. 28.8 ± 7.3 nmol/pancreas, n = 9, respectively). In line with the loss of exocrine tissue, the relative density of islets increased in CDs-HSD compared with CDr-HSD rats (1.2 ± 0.1 vs. 0.7 ± 0.1 islets/mm² pancreas area, respectively, P < 0.05, n = 5). TG content of the pancreas was significantly higher in the CDs-HSD compared with CDr-HSD rats (5.2 ± 0.9 vs. 1.2 ± 0.2 mg/g tissue, P < 0.002, n = 9). When taking into account the smaller pancreas of the CDs-HSD, the TG content of the CDs-HSD pancreas was ∼1.5-fold higher than in the CDr-HSD rat.

Levels of serum FFA were significantly higher in fasted CDs-HSD compared with CDr-HSD rats (1.442 ± 0.052 vs. 0.868 ± 0.065 mmol/l, respectively, P < 0.001).

Dense insulin immunostaining was observed in pancreatic β-cell cytoplasm, whereas dense glucagon immunostaining of the α-cells was observed as a rim at the islet periphery (Fig. 4A–D). Macrophages stained positively for interleukin (IL)-1β were found in the transition zone surrounding the islet (Fig. 4E) or as single cells in the intra-islet capillary system of CDs-HSD rats (not shown). These macrophages did not express INF-γ and tumor necrosis factor-α. Macrophages were only rarely observed in the exocrine tissue of the CDr-HSD rats (Fig. 4F). A significant increase in the number of ED1-positive macrophages was observed in the atrophic regions of the exocrine parenchyma of the CDs-HSD compared with CDr-HSD rats (2.6 ± 0.4 vs. 0.3 ± 0.1 cells/mm², respectively, P < 0.01; n = 4).

The ultrastructural analysis of the exocrine parenchyma using transmission electron microscopy (Fig. 5A–D) or the WETSEM technology (Fig. 5G and H) confirmed acinar atrophy with lipid deposits in the pancreas of the CDs-HSD rats. Nuclear heterochromatin condensation and margination typical of apoptosis were exhibited by some acinar cells of the CDs-HSD rats (Fig. 5C). No comparable signs of apoptosis or lipid deposits were found in the β-cells (Fig. 5A and B). In the fibrous transition zone between the endocrine and exocrine parenchyma, two layers of macrophages, showing ultrastructural signs of activation, including an increased cytoplasmic volume (Fig. 5D), surrounded the islet periphery and were also observed as single macrophages in the islet capillary system (not shown). The β-cells in the islet center exhibited well-preserved cellular organelles (Fig. 5E), whereas some β-cells in the islet periphery exhibited signs of cellular damage, dilated cisternae of the rough endoplasmic reticulum, and swollen mitochondria (Fig. 5F). No lipid deposits were observed in the CDr-HSD rat pancreas (Fig. 5H), in skeletal muscle, or in islets of both CDs-HSD and CDr-HSD rats (not shown). No comparable signs of apoptosis, lipid deposits, or macrophage infiltration were found in the pancreas of CDr-HSD (Supplemental Fig. 2Sa–d of the online appendix).

To test the relationships between the endocrine and exocrine compartments of the pancreas, we evaluated the dynamics of insulin release in the isolated perfused pancreas. When subjected to a square wave of glucose stimulation, the isolated pancreas of the CDr-HSD rats (n = 3) responded with biphasic insulin release (Fig. 6); the transient first phase was followed, 4–5 min later, by a gradual increase of the second phase. In contrast, the CDs-HSD rat pancreas (n = 4) expressed a very low insulin response to glucose, suggestive of a small first phase followed by a blunted second insulin phase.

As opposed to the blunted response of the perfused CDs-HSD rat pancreas to 16.7 mmol/l glucose, islets isolated from these rats responded to a similar glucose challenge with increased secretion of insulin (Fig. 7A and B). However, islets of CDs-HSD rats exhibited a left shift in insulin secretion, with significant increase already observed at 3.3 mmol/l and no additional increase between 3.3 and 16.7 mmol/l glucose. Conversely, insulin biosynthesis was shifted to the right, with increased insulin biosynthesis only apparent at 16.7 mmol/l, as opposed to 3.3 mmol/l glucose in the CDr-HSD islets. Nonetheless, similar insulin content per islet was observed in CDs-HSD and CDr-HSD rats (14.2 ± 2 and 15.1 ± 3.7 pmol/islet, respectively).

The inbred CDs-HSD rat demonstrates the importance of the cross-talk between the exocrine and endocrine compartments of the pancreas for proper β-cell function. Glucose intolerance in the Cohen diabetes–sensitive rats after high-sucrose, copper-poor diet was associated with lipid deposits and macrophage infiltration in the exocrine pancreas. Yet, lipids did not accumulate in skeletal muscle tissue or in the islets; peripheral insulin resistance or obesity were not observed in hyperglycemic CDs-HSD rats contrasting other animal models of diabetes (31–33). A pancreatic pathophysiological mechanism was therefore considered to explain the CDs-HSD diabetic phenotype.

The Cohen diabetic rat provides a good model to study GSIS. It comprises two inbred strains, one sensitive and the other resistant to the same dietary regimen which may enable deciphering the factors responsible for the reduced GSIS in the hyperglycemic CDs-HSD rat. Because a major sex difference in severity of glucose intolerance has been demonstrated, only male rats that exhibit a more severe diabetic phenotype were used in the current study (6,7). The genetic susceptibility to diet-induced glucose intolerance was observed in the present study already after 1 month of high-sucrose, copper-poor–diet feeding, 1.5 months earlier than in previous studies (6,7).

The elevated postprandial blood glucose concentrations (>12 mmol/l) and reduced insulin output observed in the hyperglycemic CDs-HSD rat were ascertained also by OGTT and IVGTT. The diminished insulin response suggested that aberrant insulin secretion is the major cause for diabetes in this model. A reduced pancreatic insulin reserve, as reported in other models of nutritional diabetes (32), could not explain the diminished insulin output in CDs-HSD rats, because insulin immunostaining, insulin content per animal and islet density were not reduced significantly in the CDs-HSD rats compared with CDr-HSD rat. Furthermore, insulin output was elicited in vivo in the hyperglycemic CDs-HSD rat by the non-nutrient secretagogues arginine and tolbutamide. This suggests the presence of sufficient insulin reserve and functional distal secretory machinery, confirming selective unresponsiveness to glucose.

Islets isolated from the pancreas of hyperglycemic CDs-HSD rats exhibited significant insulin secretion in response to 16.7 mmol/l glucose. The shifted glucose-insulin concentration-response curve may be related to the prolonged postprandial exposure of the Cohen diabetes–sensitive rat derived islets to hyperglycemia before their isolation. Similar phenomena occur in isolated islets from other diabetic models (34,35). Because insulin secretion was observed only in isolated islets and not in vivo or in the perfused pancreas experiments, we considered interdependency between the exocrine and endocrine compartment in vivo.

What could be the cause for the major discrepancy between the in vivo and in vitro insulin response to glucose? Although most animal models of type 2 diabetes do not share this phenomenon, similar observations were described in a human study performed >20 years ago (36) and recently also in the TallyHo mouse, a transgenic animal model of type 2 diabetes (37). In both studies, the reduced in vivo GSIS was not attributed to intrinsic β-cell dysfunction but was related to extrinsic factors in the surroundings of the islets. A more recent study, supporting a role for exocrine-endocrine interaction in diabetes, linked diabetes to the disrupted function of carboxyl ester lipase, the acinar cell enzyme, responsible for cholesterol esters hydrolysis (38).

In concert with these studies, several such extrinsic factors were considered as contributing to the impaired GSIS in the CDs-HSD rat. Incretin hormones, the activity of which has been shown to be attenuated in type 2 diabetes (39), were considered. However, the observation that insulin output in the CDs-HSD rat was higher in OGTT compared with the IVGTT does not support the involvement of the entero-insular hormonal axis in the in vivo secretory dysfunction.

Increased lipid deposition in the exocrine pancreas and the infiltrating macrophages have been associated with a diabetic-like phenotype in the pancreas of patients with cystic fibrosis (12), long-standing chronic pancreatitis (13), or pancreatic cancer (11) and in several transgenic mouse models (40,41). Under these conditions, fat infiltration was suggested to be associated with the development of hyperglycemia. In our study, the fasted hyperglycemic CDs-HSD rat exhibited significantly higher levels of serum FFA and increased pancreatic TG content compared with the normoglycemic CDr-HSD rat.

The specific pancreatic fat accumulation in the Cohen diabetes–sensitive rat was induced by high-sucrose, copper-poor diet, a diet low in copper and high in sucrose. Diets containing high concentrations of simple sugars were shown to affect enzymes responsible for lipolysis, resulting in fat accumulation in nonadipose tissue (42,43). Copper deficiency was also reported to be associated with abnormal glucose tolerance and pancreatic atrophy (18,19,42,43). Simple sugars, such as sucrose and fructose further augment the low copper–induced phenomenon possibly by the production of reactive oxygen species that may specifically affect GSIS (42).

Lipid deposits were observed exclusively in the exocrine parenchyma of the CDs-HSD rat. Thus, it is conceivable that the excessive fat storage in the exocrine pancreas of the CDs-HSD rat contributes to the aberrant surrounding of the β-cell and may lead to its dysfunction. On the other hand, the lack of lipids in skeletal muscle explains the high peripheral sensitivity to insulin in hyperglycemic Cohen diabetes–sensitive rats.

Previous studies have shown that macrophage infiltration is associated with lipid deposits and that these macrophages are a major source of inflammatory mediators (44–46). The infiltrating activated macrophages in the exocrine tissue of the CDs-HSD rats that expressed exclusively IL-1β could be a crucial factor in their diabetic phenotype. Other proinflammatory cytokines or inducible nitric oxide synthase (iNOS) were not expressed by the macrophages (data not shown), and T-cells were not observed in the pancreas of the hyperglycemic Cohen diabetes–sensitive rats. We believe that this may provide a plausible explanation for the fact that compared with rat models of immune-mediated diabetes (47) the Cohen diabetes–sensitive rat exhibits a mild form of type 2 diabetes characterized by a selective defect in GSIS rather than a total loss of pancreatic β-cells. This may also explain the relatively fast recovery of the GSIS when islets were isolated from the deleterious environment. Low levels of IL-1β were shown to inhibit selectively GSIS in vitro (48), and mitochondrial abnormalities similar to those observed by the ultrastructural analysis of β-cells in CDs-HSD rats were induced by IL-1β in an iNOS-independent manner (49). Thus, the hyperglycemic Cohen diabetes–sensitive rat may represent a model of a mild cytokine-mediated diabetic syndrome.

In conclusion, glucose intolerance in the diabetic Cohen diabetes–sensitive rat is a result of its genetic susceptibility to a copper-poor, sucrose-rich diet. A blunted β-cell response to glucose in vivo is responsible for the diabetic phenotype. The association of the in vivo secretory defect with lipid accumulation in the exocrine pancreas and the observation of IL-1β–positive activated macrophages in the islet vicinity suggest that changes in the islet microenvironment are the culprit in the in vivo malfunction of the β-cell.

S.W.-Z. has received grants from the AM Cohen Foundation for the Advancement of Research of the Cohen Diabetic Rat. N.K. has received a grant from the Israel Science Foundation. This work has received support from the Russell Berrie Foundation and D-Cure, Diabetes Care in Israel.

We thank Limor Chen, Carol Levy, Elena Dvorkin, and Yaffa Ariav for assistance in performing some of these studies. We thank Debbie Anaby from QuantomiX for performing the WETSEM analysis of the pancreases and muscles.