Chronic Inhibition of Dipeptidyl Peptidase-4 With a Sitagliptin Analog Preserves Pancreatic β-Cell Mass and Function in a Rodent Model of Type 2 Diabetes

The DPP-4 inhibitor used in this study was a des-fluoro analog of sitagliptin, 7-[(3R)-3-amino-1-oxo-4-(2,5-difluorophenyl)butyl]-5,6,7,8-tetrahydro-3-(trifluoromethyl)-1,2,4-triazolo[4,3-a]pyrazine l-tartaric acid salt (compound 25 in ref. 25), and is referred to herein as des-fluoro-sitagliptin. Des-fluoro-sitagliptin was prepared by Process Research, Merck Research Labs (Rahway, NJ). Glipizide, STZ, and other chemicals were purchased from Sigma Chemical (St. Louis, MO).

All animal procedures were performed in accordance with the guidelines of the institutional animal care and use committee of Merck. Mice were housed eight per cage and allowed access to diet and autoclaved water. Animal housing rooms were maintained at a constant room temperature (25°C) in a 12-h light (7:00 a.m.)/dark (7:00 p.m.) cycle.

Four-week-old male ICR mice were purchased from Taconic Farm (Germantown, NY) and placed on the HFD D12492 (Research Diets, New Brunswick, NJ), in which 60% of kilocalories is from fat. After 3 weeks of HFD feeding, the mice were injected once with low-dose STZ (intraperitoneal at 90–100 mg/kg) to induce partial insulin deficiency. Three weeks after STZ injection, the majority of HFD/STZ-treated mice displayed hyperglycemia, insulin resistance, and glucose intolerance as previously reported (36). At ∼10 weeks of age (and 6 weeks of HFD feeding), animals with similar degrees of hyperglycemia and body weight were randomly divided to various vehicle or compound treatment groups. The normal diet–fed mice were used as nondiabetic controls.

Des-fluoro-sitagliptin was administrated orally by premixing with the HFD D12492 at 0.1, 0.4, and 1.1% (wt/wt; the mixing and repelleting was performed by Research Diets). Glipizide was also dosed as admixture to HFD at 0.02%. The target doses were 43, 208, and 576 mg/kg (milligrams drug per kilograms body weight) for des-fluoro-sitagliptin and 20 mg/kg for glipizide.

Postprandial plasma glucose, body weight, and food consumption were monitored weekly. Plasma HbA_1c (A1C) level was measured using a Micromat II test kit from Bio-Rad Laboratories. Six-hour fasting glucose levels were also monitored. Hepatic triglyceride levels were measured as described (28). Plasma drug levels were measured by liquid chromatography/tandem mass spectrometry; plasma DPP-4 activities and GLP-1_intact levels were measured as previously described (25).

To study the acute effect of glipizide on glucose lowering, HFD/STZ mice were given oral administration of vehicle (0.5% methylcellulose) or 10 mg/kg of glipizide. Glucose levels were monitored at 0, 1, 2, 3, and 4 h after dosing. Age-matched ICR mice without HFD/STZ treatment were used as controls.

Mice were fasted for 4 h before glucose tolerance tests. Oral glucose load was administered at 2 g/kg of body weight. Glucose levels were measured from tail bleeds with a glucometer (Lifescan, Milpitas, CA) at specified time points after glucose administration.

Measurement of islet cell mass was performed using modifications of previously described methodologies (9,12,29,30). In brief, pancreatic tissue was collected at necropsy and immediately frozen in Tissue Freezing Medium (Electron Microscopy Sciences, Hatfield, PA)–filled tissue molds (for insulin/glucagon labeling). Cryostat sections were immunolabeled with rabbit anti-glucagon followed by guinea pig anti-insulin (or a combination of anti-insulin, anti-glucagon, and anti-somatostatin antibodies). Digital images were captured in a Zeiss Axioplan II microscope equipped with a Slidebook Image analysis workstation (Intelligent Imaging Innovations, Denver, CO).

The insulin-positive β-cell–to–total islet area and glucagon positive α-cell–to–total islet area ratios were calculated from digitized images captured through the 20× objective using the Slidebook software. Images of 10 randomly chosen islets were captured from each section. Immunolabeled sections were prepared from four animals per treatment group.

Analyses of all immunohistochemical images were carried out twice by two separate operators who were blinded to the treatment group.

Pancreatic islets of Langerhans were isolated from the pancreata by collagenase digestion and discontinuous Ficoll gradient separation, a modification of the original method of Lacy and Kostianovsky (31). The islets were cultured 2 h in RPMI-1640 medium (11 mmol/l glucose) before GSIS assay.

To measure GSIS, islets were incubated with 2, 8, or 16 mmol/l glucose or 2 mmol/l glucose + 30 mmol/l KCl to measure maximum insulin secretion. Insulin was measured in aliquots of the incubation buffer by enzyme-linked immunosorbent assay with a commercial kit (ALPCO Diagnostics, Windham, NH). Insulin content was measured after acid ethanol (0.18 mol/l HCl in 70% ethanol) extraction as described (32).

The whole pancreas was dissected free from fat and other nonpancreas tissue immediately after the mouse was killed. After the measurement of the wet weight, the pancreas was extracted with the acid ethanol method, and insulin concentration was determined.

All data are expressed as means ± SE. Statistical analysis was conducted by using Student’s t test. Statistical significance was defined as P < 0.05.

To generate a nongenetic rodent model mimicking human type 2 diabetes with a combination of insulin resistance and insulin deficiency, ICR mice were fed an HFD for 3 weeks and then injected with a single low dose of STZ followed by continued HFD feeding for an additional 3 weeks (33). As shown in Table 1, HFD or STZ injection alone did not significantly affect blood glucose, whereas the combination of HFD and STZ treatment led to frank hyperglycemia. The low dose of STZ caused an ∼40% reduction of plasma insulin levels, which was not sufficient to cause hyperglycemia in mice fed with normal diet. In contrast, this level of hypoinsulinemia precipitated hyperglycemia in the face of insulin resistance induced by the HFD. The mean body weight of HFD/STZ mice was comparable to that of normal diet–fed control animals, but their plasma triglyceride and free fatty acid (FFA) levels were significantly elevated. The HFD/STZ mouse model thus manifests hyperglycemia and hyperlipidemia associated with insulin resistance and impaired insulin secretion. We therefore choose it as a model of type 2 diabetes to study the effects of chronic DPP-4 inhibition.

DPP-4 inhibitors improve glycemic control in animal models and in type 2 diabetic subjects, but the long-term consequences of DPP-4 inhibition on pancreatic β-cell function and β-cell mass are not fully characterized. We therefore performed an 11-week study to test the effect of the selective DPP-4 inhibitor des-fluoro-sitagliptin on glycemic control and pancreatic islet morphology in the HFD/STZ mouse model. Des-fluoro-sitagliptin is a potent and selective DPP-4 inhibitor with half-maximal inhibitory concentration values of 27 and 97 nmol/l against human and mouse DPP-4, respectively (25). The potency, selectivity, and pharmacokinetic properties of this compound are virtually identical to those of sitagliptin. HFD/STZ mice maintained on a standard light-dark cycle were treated with the compound as admixture to the diet at 0.1, 0.4, and 1.1% (corresponding to 43, 208, and 576 mg/kg). Mean plasma peak/trough drug levels (estimated by measurements at 8:00 a.m./4:00 p.m.) were 0.7/0.2, 3.5/1.0, and 17.6/1.3 μmol/l at 0.1, 0.4, and 1.1% des-fluoro-sitagliptin, respectively. The high doses of des-fluoro-sitagliptin needed to achieve these efficacious drug levels reflect the short half-life of the compound in mice (1–2 h). At the high dose used in the study, >95% inhibition of DPP-4 in the plasma was achieved at the trough drug levels (>1 umol/l) measured at 4:00 p.m. The animals ingested the inhibitor while feeding during the dark phase. Postprandial glucose levels were measured 1 h after the beginning of the light phase.

As illustrated in Fig. 1A–C, significant and dose-dependent reductions of postprandial and 6-h fasting blood glucose, as well as of A1C, were observed over the course of the study. Body weight and food intake were not significantly different between vehicle- and des-fluoro-sitagliptin–treated groups (data not shown). Vehicle-treated HFD/STZ mice also manifested hepatomegaly and hepatic steatosis relative to normal diet–fed control mice at the termination of the study. Treatment with des-fluoro-sitagliptin mitigated these abnormalities and resulted in significant and dose-dependent reductions in liver weight (Fig. 2A), hepatic triglyceride content (Fig. 2B), circulating triglycerides (Fig. 2C), and FFA levels (Fig. 2D) relative to vehicle. These results suggest that chronic efficacy of DPP-4 inhibitors may extend beyond improved glycemic control to include beneficial effects on other metabolic disturbances associated with type 2 diabetes, including hyperlipidemia and hepatic steatosis.

At termination of the 11-week des-fluoro-sitagliptin treatment, the pancreas of each animal was cryopreserved, and multiple cryostat sections were obtained from each sample for measurement of β-cell mass and characterization of islet morphology.

Dramatic differences in islet architecture and islet cell composition were observed by immunohistochemistry between nondiabetic control mice and various HFD/STZ groups at the end of the study (Fig. 3). As expected, islets of normal diet–fed control mice comprised a large insulin-positive β-cell core surrounded by a mantle of glucagon-positive α-cells. In contrast, islets from vehicle-treated HFD/STZ diabetic mice contained many more glucagon-positive α-cells, which infiltrated the entire islet including the central core. The 11-week treatment with increasing doses of des-fluoro-sitagliptin significantly reduced the number of α-cells in the islet core and restored the normal β-cell/α-cell distribution pattern in islets.

We also examined low-power digital images of whole pancreata sections after staining with a combination of anti-insulin, anti-glucagon, and anti-somatostation antibodies. These sections generally contained 200–300 islets and thus provided a full view of the islet area in each animal. Analysis of these images demonstrated that vehicle-treated HFD/STZ mice had ∼50% reduction of total β-cell mass relative to normal diet–fed nondiabetic controls, as measured by percent β-cell area/total pancreas area (Fig. 4A). Treatment with the DPP-4 inhibitor restored β-cell mass but did not expand it beyond that of normal mice.

Insulin-positive β-cell–to–total islet area and glucagon-positive α-cell–to–total islet area ratios were calculated from acquired digitized images. As shown in Fig. 4B, the ratio of insulin-positive β-cells to total islet area was reduced from 0.62 ± 0.06 in nondiabetic control mice to 0.44 ± 0.10 in diabetic mice treated with vehicle (P < 0.001, n = 40 islets). Treatment of the diabetic mice with des-fluoro-sitagliptin dose dependently increased the insulin-positive β-cell–to–total islet area ratio with a complete normalization of the ratio in the two higher-dose–treated groups. On the other hand, there was an elevation in the ratio of glucagon-positive α-cell to total islet area in diabetic mice compared with the nondiabetic control mice (Fig. 4C). Treatment of the diabetic mice with des-fluoro-sitagliptin reduced the glucagon-positive α-cell–to–total islet area ratio to the normal range. Consistent with these morphological findings, we also observed a slight but significant increase in circulating glucagon levels in vehicle-treated HFD/STZ mice (60.8 ± 5.6 pg/ml, n = 18) compared with nornal diet–fed control mice (34.6 ± 3.9 pg/ml, n = 18), which was also alleviated by treatment with the DPP-4 inhibitor at the high dose (49.4 ± 5.0 pg/ml, n = 18).

The sulfonylurea class of ATP-sensitive potassium channel blockers such as glipizide are commonly used in the treatment of type 2 diabetes as monotherapy or in combination with other agents (34–37). Patients on sulfonylurea therapy often convert to insulin therapy progressively due to unsatifactory glycemic control (38,39). Following a single oral dose of glipizide at 10 mg/kg, significant reductions in blood glucose levels were observed in diabetic HFD/STZ mice as well as in control nondiabetic mice (Fig. 5A). When islets isolated from control and HFD/STZ mice were incubated with 2 mmol/l glucose in the presence of glipizide, glipizide was effective in stimulating insulin secretion in a dose-dependent manner (Fig. 5B). As expected, islets from nondiabetic control mice responded better to glipizide or 16 mmol/l glucose for insulin secretion (Fig. 5B). Taken together, these data demonstrate that the HFD/STZ mice are responsive to glipizide following acute treatment.

We then compared the long-term efficacy of des-fluoro-sitagliptin and glipizide on glycemic control in HFD/STZ mice. Diabetic mice were treated with either 576 mg/kg des-fluoro-sitagliptin (administered as 1.1% of diet) or 20 mg/kg glipizide (0.02% of diet) for 10 weeks. Plasma peak/trough drug levels achieved were 20.1/1.4 μmol/l for des-fluoro-sitagliptin and 28.2/13.3 μmol/l for glipizide in this study.

As illustrated in Fig. 5C, significant corrections of postprandial hyperglycemia were observed in mice following a 10-week treatment with des-fluoro-sitagliptin. Glipizide administration in the same paradigm resulted in postprandial glucose lowering only in the first 4 weeks, but its efficacy diminished afterward possibly due to exhaustion of insulin-producing capacity of β-cells. Furthermore, elevated A1C levels (assessed at the termination of the 10-week treatment) were significantly alleviated only by des-fluoro-sitagliptin but not by glipizide treatment (Table 2). To further evaluate glycemic control among the various groups, we performed an oral glucose tolerance test study after 8 weeks of compound treatment. Vehicle-treated diabetic mice manifested significantly elevated glucose excursions following oral glucose challenge compared with nondiabetic control mice. The glucose excursion was significantly reduced in diabetic mice treated with des-fluoro-sitagliptin (Fig. 5D) but not with glipizide. Furthermore, treatment with des-fluoro-sitagliptin resulted in an elevation of circulating GLP-1_intact and a reduction in plasma glucagon levels, whereas glipizide treatment had no effects on these hormones following the 10-week treatment (Table 2). Thus, the DPP-4 inhibitor exhibits superior antihyperglycemic efficacy compared with sulfonylurea in this animal model.

Contrasting effects of des-fluoro-sitagliptin and glipizide on plasma triglyceride and FFA levels were also observed in the present study. Elevations in plasma triglyceride and FFA levels in HFD/STZ mice were completely prevented by the 10-week treatment with des-fluoro-sitagliptin, whereas glipizide administration over the same period had no effect on these parameters (Table 2).

Cyrostat pancreata sections of animals from the study were analyzed by anti-insulin and anti-glucagon double immunolabeling to evaluate islet morphology. As illustrated in Fig. 6A and B, islets from vehicle-treated diabetic mice had markedly reduced β-cell–to–islet area ratio compared with islets from nondiabetic control mice. Consistent with the study shown in Fig. 3, treatment with des-fluoro-sitagliptin corrected these abnormalies in islet β-cell composition and distribution in the HFD/STZ mice. In contrast, glipizide failed to correct the reduced β-cell–to–islet area ratio in HFD/STZ mice. The contrasting effects of des-fluoro-sitagliptin and glipizide on β-cell mass were also clearly reflected in the measurement of pancreatic insulin content. As shown in Fig. 6C, vehicle-treated HFD/STZ diabetic mice suffered severe depletion of pancreatic insulin content compared with normal diet–fed normal control mice. Treatment of the diabetic animals with des-fluoro-sitagliptin, but not glipizide, partially restored pancreatic insulin content.

The effects of chronic inhibition of DPP-4 on the function of pancreatic islets were also evaluated at the end of the chronic study using isolated islets from animals under different treatments. Significant reductions in insulin content (Fig. 7A) and maximum releasable insulin (stimulated by 30 mmol/l KCl) in islets (Fig. 7B) from vehicle- and glipizide-treated diabetic mice were observed compared with nondiabetic control mice. Treatment with des-fluoro-sitagliptin led to significant increases in islet insulin content and the maximum insulin secretion capacity in HFD/STZ mice. GSIS was also assessed by incubating the islets in media containing 2, 8, or 16 mmol/l glucose. As shown in Fig. 7C, islets from vehicle-treated diabetic mice exhibited significantly impaired GSIS relative to islets from normal diet–fed control mice. Islets from des-fluoro-sitagliptin–treated mice exhibited significantly improved insulin secretory responses to 8 and 16 mmol/l glucose, whereas no improvement in GSIS was observed in islets from glipizide-treated mice. These results indicate that chronic treatment with the DPP-4 inhibitor, but not with glipizide, significantly improves islet morphology and function in this model.

A key challenge in the management of type 2 diabetes is the gradual deterioration of glycemic control over time, which is believed to be linked to the progressive loss of β-cell mass and function (2,4). A critical, unmet medical need is therefore to alter the course of the disease and to achieve durability of efficacy. Therapies that maintain the capacity of the β-cell to synthesize and release insulin are most likely to fulfill this unmet need. An agent that increases β-cell mass and function over time could inhibit the natural progression of type 2 diabetes, leading to delayed deterioration of glycemic control, and a decrease in the need for combination therapy to lower blood glucose. DPP-4 inhibitors are a new therapeutic approach to diabetes that act, at least in part, by increasing GSIS (13–15). This effect is believed to be mediated primarily via stabilization of the incretin hormones including GLP-1.

In the current study, we report that chronic treatment in a nongenetic mouse model of type 2 diabetes with a potent and selective DPP-4 inhibitor resulted in significant improvement in glycemic control and metabolic profile. The improved glucose homeostasis was associated with dose-dependent increases in β-cell mass, increased ratios of insulin-positive β-cell to total islet area, and the restoration of normal islet architecture (β-/α-cell distribution pattern) (Figs. 3 and 4). These findings highlight the potential utility of DPP-4 inhibitors in tackling the underlying pathogenic cause of type 2 diabetes, namely the progressive loss of pancreatic β-cell mass. The therapeutic potential of DPP-4 inhibitors versus conventional insulin secretagogues like the sulfonylureas is highlighted by our comparison of chronic DPP-4 inhibitor administration with glipizide treatment in the same model. Glipizide is a sulfonylurea that is in current clinical use for the treatment of type 2 diabetes (34–36). In the present study, glipizide was equally effective in reducing blood glucose levels in HFD/STZ mice as des-fluoro-sitagliptin during the first few weeks of treatment, but it gradually lost its efficacy on blood glucose late in the study. Thus, while des-fluoro-sitagliptin was efficacious in decreasing A1C in diabetic mice following treatment for 10 weeks, glipizide was ineffective in lowering A1C (Fig. 5). With long-term use, there is a progressive decrease in the effectiveness of sulfonylureas resulting from a gradual reduction in and eventual exhaustion of insulin-producing capacity of the pancreatic β-cell. When islets were isolated from control and treated animals and evaluated ex vivo, it was apparent that only des-fluoro-sitagliptin, but not glipizide, was effective in replenishing severely depleted islet insulin content and improving islet β-cell insulin secretory function measured by glucose- and KCl-stimulated insulin secretion (Fig. 7).

It is possible that the observed beneficial effects of the DPP-4 inhibitor on β-cells could simply be a result of improvements in glycemia control. However, it is highly plausible that the beneficial effects of des-fluoro-sitagliptin on glycemic regulation and β-cell mass are at least partially mediated via increased GLP-1 and GIP signaling. This hypothesis would be consistent with previous reports that treatment with DPP-4 inhibitors increases circulating GLP-1 levels and stimulates β-cell survival and proliferation in STZ-induced diabetic rats (26,27). In addition, we and others have shown that mice deficient in DPP-4 (CD26^−/−) have elevated levels of circulating GLP-1 and GIP and are resistant to STZ-induced β-cell destruction (40,41). Furthermore, GLP-1 treatment has been shown to prevent STZ-induced apoptosis of a β-cell line (INS-1) in vitro in a dose-dependent manner (37). Similarly, GIP has also been reported to stimulate β-cell proliferation (37,42). Results from a recent study using mice lacking both GLP-1 and GIP receptors clearly indicate that both of these incretins are mediators of the acute effects of DPP-4 inhibitors on glycemic control (43). A recent study reported by Ahren and Hughes (44) indicated that inhibition of DPP-4 by valine pyrrolidide in mice resulted in augmentation of the bioactivity of exogenously administered peptide hormones, including not only GLP-1 and GIP but also PACAP and GRP. The potential effects of PACAP and GRP on β-cell apoptosis and proliferation may warrant further investigation. Detailed studies on the mechanisms of β- and α-cell mass regulation and the regulation of insulin production and insulin secretion in this regard will be the subject of follow-up studies.

While both glipizide and des-fluoro-sitagliptin were effective in reducing postprandial hyperglycemia, only the DPP-4 inhibitor corrected disturbances in lipid metabolism in this study. Similar lipid-lowering effects have also been observed with another DPP-4 inhibitor in STZ-induced diabetic rats (26) and ZDF rats (45). It is yet to be determined whether the lipid-lowering effects of DPP-4 inhibitors in preclinical studies are mediated by the enhanced action of incretin hormones. Nevertheless, elevated FFA levels have been demonstrated to be detrimental to the pancreatic β-cell (the so-called lipotoxicity) (46–48). The normalization of plasma lipids may contribute in part to the beneficial effects of des-fluoro-sitagliptin on β-cell function and islet morphology observed herein.

In conclusion, results from the current study show that des-fluoro-sitagliptin is efficacious in improving glucose homeostasis and metabolic profile in a mouse model of type 2 diabetes with defects in insulin sensitivity and secretion. The beneficial effect of the DPP-4 inhibitor on glycemia control is clearly associated with significantly increased β-cell mass and function. Furthermore, chronic DPP-4 inhibition with des-fluoro-sitagliptin demonstrated superior glucose-lowering efficacy and β-cell–preserving effects compared with glipizide. These findings suggest that the beneficial effects of DPP-4 inhibitors could extend beyond glycemic regulation via modulation of insulin secretion to include a potential reduction in β-cell failure commonly observed in diabetic patients. The potential for DPP-4 inhibitors to alter the course of diabetes as a progressive disease warrants further studies in the clinic.

We thank Kathy Lyons for measuring drug levels in plasma samples in the studies and Dr. Ann Weber for helpful discussions.